Dissertations / Theses on the topic '3D Networks'
To see the other types of publications on this topic, follow the link: 3D Networks.
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic '3D Networks.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
1
Stigeborn, Patrik. "Generating 3D-objects using neural networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-230668.
Full textAbstract:
Enabling a 2D- to 3D-reconstruction is an interesting future service for Mutate AB, where this thesis is conducted. Convolutional neural networks (CNNs) is examined in different aspects, in order to give a realistic perception of what this technology is capable of. The task conducted, is the creation of a CNN that can be used to predict how an object from a 2D image would look in 3D. The main areas that this CNN is optimized for are Quality, Speed, and Simplicity. Where Quality is the output resolution of the 3D object, Speed is measured by the number of seconds it takes to complete a reconstruction, and Simplicity is achieved by using machine learning (ML). Enabling this could potentially ease the creation of 3D games and make the development faster. The chosen solution is to use two CNNs. The first CNN is using convolution to extract features from an input image. The second CNN is using transpose convolution to create a prediction of how the object would look in 3D, from the features extracted by the first neural network. This thesis is using an empirical development approach to reach an optimal solution for the CNN structure and its hyperparameters. The 3D-reconstruction is inspired by a sculpting process, meaning that the reconstruction starts with a low resolution and improves it iteratively. The result shows that the quality gained from each iteration grows exponentially whilst the increased time grows a lot less. Thereof, the conclusion is that the trade-off between speed and quality is in our favor. However, when looking at commercializing this technology or deploy it in a professional environment, it is still too slow to generate high resolution output. Also, in this case, the CNN is fragile when there are a lot of unrecognized shapes in the input image.
2
Costa, Breno Jacinto Duarte da. "3D Routing with Context Awareness." Universidade Federal de Pernambuco, 2009. https://repositorio.ufpe.br/handle/123456789/1771.
Full textAbstract:
Made available in DSpace on 2014-06-12T15:52:15Z (GMT). No. of bitstreams: 1
license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5)
Previous issue date: 2009Conselho Nacional de Desenvolvimento Científico e Tecnológico
O surgimento de interfaces de rede sem-fio de baixo custo no mercado e o crescimento na demanda por dispositivos móveis (como Smartphones, PDAs, Internet Tablets e Laptops) permitiram a criação de cenários onde serviços de rede para usuários móveis possam existir sem nenhuma infra-estrutrutura pré-configurada. No entanto, a interoperabilidade entre tais redes, que são dinâmicas e heterogêneas, é atualmente objeto de pesquisa. Várias pesquisas na área de redes ad hoc sem-fio tem focado em uma única tecnologia sem-fio, baseada no padrão IEEE 802.11, onde os nós da rede são vistos de maneira plana (2D), ou seja, como elementos homogêneos, identificados apenas por endereços IP, não levando em consideração seus perfis de hardware e tecnologias de rede. Desta forma, pesquisas envolvendo mais de uma tecnologia de rede encontram-se em estágios iniciais. Novas propostas são necessárias para estes cenários, que são cada vez mais comuns, envolvendo múltiplos dispositivos com múltiplas interfaces de rede (multi-homed). Este trabalho propõe o protocolo de roteamento 3D, direcionado a cenários onde há heterogeneidade de dispositivos e tecnologias de rede. O objetivo do protocolo de roteamento proposto é prover mecanismos para a interoperabilidade de redes ad hoc heterogêneas, considerando outra dimensão de informações, aqui denominada de terceira dimensão (3D), que consiste em agregar mais informações, como informações de contexto, recursos dos dispositivos e interfaces de rede, ao processo de roteamento. Para isto, o protocolo considera os seguintes aspectos fundamentais: o processo de bootstrapping da rede heterogênea e dos nós, a construção e disseminação de informações de ciência de contexto entre os nós, e a atribuição de papéis específicos para determinados nós da rede. A avaliação do protocolo é feita através de experimentos em um test-bed real, utilizando um protótipo da implementação do protocolo, num cenário composto de dispositivos móveis como Smartphones OpenMoko, Internet Tablets N810 da Nokia e Laptops, possuindo tecnologias Bluetooth e 802.11, executando versões embarcadas do sistema operacional Linux
3
Zhao, Yao. "Autonomous Localization in 3D Surface Wireless Sensor Networks." Thesis, University of Louisiana at Lafayette, 2014. http://pqdtopen.proquest.com/#viewpdf?dispub=3622968.
Full textAbstract:
Location awareness is imperative for a variety of sensing applications and network operations. Although a diversity of GPS-less and GPS-free solutions have been developed recently for autonomous localization in wireless sensor networks, they primarily target at 2D planar or 3D volumetric settings. There exists unique and fundamental hardness to extend them to 3D surfaces.
The contributions of this work are twofold. First, it proposes a theoretically-proven algorithm for the 3D surface localization problem. Seeing the challenges to localize general 3D surface networks and the solvability of the localization problem on single-value (SV) surface, this work proposes the cut-and-sew algorithm that takes a divide-and-conquer approach by partitioning a general 3D surface network into SV patches, which are localized individually and then merged into a unified coordinates system. The algorithm is optimized by discovering the minimum SV partition, an optimal partition that creates a minimum set of SV patches.
Second, it develops practically-viable solutions for real-world sensor network settings where the inputs are often noisy. The proposed algorithm is implemented and evaluated via simulations and experiments in an indoor testbed. The results demonstrate that the proposed cut-and-sew algorithm achieves perfect 100% localization rate and the desired robustness against measurement errors.
4
Cronje, Frans. "Human action recognition with 3D convolutional neural networks." Master's thesis, University of Cape Town, 2015. http://hdl.handle.net/11427/15482.
Full textAbstract:
Convolutional neural networks (CNNs) adapt the regular fully-connected neural network (NN) algorithm to facilitate image classification. Recently, CNNs have been demonstrated to provide superior performance across numerous image classification databases including large natural images (Krizhevsky et al., 2012). Furthermore, CNNs are more readily transferable between different image classification problems when compared to common alternatives. The extension of CNNs to video classification is simple and the rationale behind the components of the model are still applicable due to the similarity between image and video data. Previous CNNs have demonstrated good performance upon video datasets, however have not employed methods that have been recently developed and attributed improvements in image classification networks. The purpose of this research to build a CNN model that includes recently developed elements to present a human action recognition model which is up-to-date with current trends in CNNs and current hardware. Focus is applied to ensemble models and methods such as the Dropout technique, developed by Hinton et al. (2012) to reduce overfitting, and learning rate adaptation techniques. The KTH human action dataset is used to assess the CNN model, which, as a widely used benchmark dataset, facilitates the comparison between previous work performed in the literature. Three CNNs are built and trained to provide insight into design choices as well as allow the construction of an ensemble model. The final ensemble model achieved comparative performance to previous CNNs trained upon the KTH data.
While the inclusion of new methods to the CNN model did not result in an improvement on previous models, the competitive result provides an alternative combination of architecture and components to other CNN models.
5
Nihlén, Ramström Max. "Sketch to 3D Model using Generative Query Networks." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-251507.
Full textAbstract:
For digital artists and animators, translating an idea from a rough sketch to a 3D model is a time consuming process requiring a plethora of different software. In this work, a Generative Model which can directly generate images of 3D models from arbitrary view points by observing sketched 2D images is presented. The model is based on Generative Query Networks and two different generative models were tested for generating new images, the first a Variational Auto Encoder and the second a Generative Adversarial Network. The model learns to produce new images from any queried view point allowing it to perform so called mental rotation of an object as if a 3D model had been generated. A paired dataset containing images of 3D models, the view point from where each image is captured and corresponding sketch versions was created in order to train the model. It was found that the Variational Auto Encoder could create plausible images from as little as a single sketch while the Generative Adversarial Network failed to correctly condition on the given sketches.För digitala artister och animatörer är processen att gå ifrån en idé i form av en sketch till en färdig 3D-modell tidskrävande och sträcker sig över en mängd olika mjukvaror. Detta arbete presenterar en Generativ Modell som direkt kan generera bilder av en 3D-modell ifrån sketchade bilder i 2D. Modellen är baserad på Generative Query Networks och två olika Generativa Modeller testades för att generera nya bilder, den första en Variational Auto Encoder och den andra en Generative Adversarial Network. Modellen lär sig att skapa nya bilder ifrån godtyckliga synvinklar vilket tillåter den att utföra så kallad mental rotation av ett objekt på samma sätt som om en 3D-modell hade genererats. För att kunna träna modellen skapades ett dataset där bilder sparades både i ursprungs- samt i sketchform tillsammans med synvinklarna där bilderna tagits ifrån. Modellen som använde sig av en Variational Auto Encoder visade sig kunna generera trovärdiga bilder efter att endast ha observerat en sketch medan modellen som använde ett Generative Adversarial Network misslyckades med att betinga de genererade bilderna på de sketcher den observerat.
6
Mohib, Hamdullah. "End-to-end 3D video communication over heterogeneous networks." Thesis, Brunel University, 2014. http://bura.brunel.ac.uk/handle/2438/8293.
Full textAbstract:
Three-dimensional technology, more commonly referred to as 3D technology, has revolutionised many fields including entertainment, medicine, and communications to name a few. In addition to 3D films, games, and sports channels, 3D perception has made tele-medicine a reality. By the year 2015, 30% of the all HD panels at home will be 3D enabled, predicted by consumer electronics manufacturers. Stereoscopic cameras, a comparatively mature technology compared to other 3D systems, are now being used by ordinary citizens to produce 3D content and share at a click of a button just like they do with the 2D counterparts via sites like YouTube. But technical challenges still exist, including with autostereoscopic multiview displays. 3D content requires many complex considerations--including how to represent it, and deciphering what is the best compression format--when considering transmission or storage, because of its increased amount of data. Any decision must be taken in the light of the available bandwidth or storage capacity, quality and user expectations. Free viewpoint navigation also remains partly unsolved. The most pressing issue getting in the way of widespread uptake of consumer 3D systems is the ability to deliver 3D content to heterogeneous consumer displays over the heterogeneous networks. Optimising 3D video communication solutions must consider the entire pipeline, starting with optimisation at the video source to the end display and transmission optimisation. Multi-view offers the most compelling solution for 3D videos with motion parallax and freedom from wearing headgear for 3D video perception. Optimising multi-view video for delivery and display could increase the demand for true 3D in the consumer market. This thesis focuses on an end-to-end quality optimisation in 3D video communication/transmission, offering solutions for optimisation at the compression, transmission, and decoder levels.
7
Birgersson, Anna, and Klara Hellgren. "Texture Enhancement in 3D Maps using Generative Adversarial Networks." Thesis, Linköpings universitet, Datorseende, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-162446.
Full textAbstract:
In this thesis we investigate the use of GANs for texture enhancement. To achievethis, we have studied if synthetic satellite images generated by GANs will improvethe texture in satellite-based 3D maps. We investigate two GANs; SRGAN and pix2pix. SRGAN increases the pixelresolution of the satellite images by generating upsampled images from low resolutionimages. As for pip2pix, the GAN performs image-to-image translation bytranslating a source image to a target image, without changing the pixel resolution. We trained the GANs in two different approaches, named SAT-to-AER andSAT-to-AER-3D, where SAT, AER and AER-3D are different datasets provided bythe company Vricon. In the first approach, aerial images were used as groundtruth and in the second approach, rendered images from an aerial-based 3D mapwere used as ground truth. The procedure of enhancing the texture in a satellite-based 3D map was dividedin two steps; the generation of synthetic satellite images and the re-texturingof the 3D map. Synthetic satellite images generated by two SRGAN models andone pix2pix model were used for the re-texturing. The best results were presentedusing SRGAN in the SAT-to-AER approach, in where the re-textured 3Dmap had enhanced structures and an increased perceived quality. SRGAN alsopresented a good result in the SAT-to-AER-3D approach, where the re-textured3D map had changed color distribution and the road markers were easier to distinguishfrom the ground. The images generated by the pix2pix model presentedthe worst result. As for the SAT-to-AER approach, even though the syntheticsatellite images generated by pix2pix were somewhat enhanced and containedless noise, they had no significant impact in the re-texturing. In the SAT-to-AER-3D approach, none of the investigated models based on the pix2pix frameworkpresented any successful results. We concluded that GANs can be used as a texture enhancer using both aerialimages and images rendered from an aerial-based 3D map as ground truth. Theuse of GANs as a texture enhancer have great potential and have several interestingareas for future works.
8
Nguyen, Thu Duc. "System support for distributed 3D real-time rendering on commodity clusters /." Thesis, Connect to this title online; UW restricted, 1999. http://hdl.handle.net/1773/7018.
Full text
9
Ahmad, Waqar. "Core Switching Noise for On-Chip 3D Power Distribution Networks." Doctoral thesis, KTH, Elektroniksystem, 2012. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-103566.
Full textAbstract:
Reducing the interconnect size with each technology node and increasing speed with each generation increases IR-drop and Ldi/dt noise. In addition to this, the drive for more integration increases the average current requirement for modern ULSI design. Simultaneous switching of core logic blocks and I/O drivers produces large current transients due to power distribution network parasitics at high clock frequency. The current transients are injected into the power distribution planes thereby inducing noise in the supply voltage. The part of the noise that is caused by switching of the internal logic load is core switching noise. The core logic switches at much higher speed than driver speed whereas the package inductance is less than the on-chip inductance in modern BGA packages. The core switching noise is currently gaining more attention for three-dimensional integrated circuits where on-chip inductance is much higher than the board and package inductance due to smaller board, and package. The switching noise of the driver is smaller than the core switching noise due to small driver size and reduced capacitance associated with short on-board wires for three-dimensional integrated circuits. The load increases with the addition of each die. The power distribution TSV pairs to supply each extra die also introduce additional parasitic. The core switching noise may propagate through substrate and consequently through interconnecting TSVs to different dies in heterogeneous integrated system. Core switching noise may lead to decreased device drive capability, increased gate delays, logic errors, and reduced noise margins. The actual behavior of the on-chip load is not well known in the beginning of the design cycle whereas altering the design during later stages is not cost effective. The size of a three-dimensional power distribution network may reach billions of nodes with the addition of dies in a vertical stack. The traditional tools may run out of time and memory during simulation of a three-dimensional power distribution network whereas, the CAD tools for the analysis of 3D power distribution network are in the process of evolution. Compact mathematical models for the estimation of core switching noise are necessary in order to overcome the power integrity challenges associated with the 3D power distribution network design. This thesis presents three different mathematical models to estimate core switching noise for 3D stacked power distribution networks. A time-domain-based mathematical model for the estimation of design parameters of a power distribution TSV pair is also proposed. Design guidelines for the estimation of optimum decoupling capacitance based on flat output impedance are also proposed for each stage of the vertical chain of power distribution TSV pairs. A mathematical model for tradeoff between TSV resistance and amount of decoupling capacitance on each DRAM die is proposed for a 3D-DRAM-Over-Logic system. The models are developed by following a three step approach: 1) design physical model, 2) convert it to equivalent electrical model, and 3) formulate the mathematical model based on the electrical model. The accuracy, speed and memory requirement of the proposed mathematical model is compared with equivalent Ansoft Nexxim models.QC 20121015
10
Nordhus, Lars Espen Strand. "Ray Tracing for Simulation of Wireless Networks in 3D Scenes." Thesis, Norges teknisk-naturvitenskapelige universitet, Institutt for datateknikk og informasjonsvitenskap, 2013. http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-23002.
Full textAbstract:
Simulating WIFI and other similar radio waves in real-time environments has a tremendous potential, and is a hot topic in modern computer science. The Norwegian Road Authority in Trondheim (Norway) has created a physical test road for simulating future road to vehicle communications in a realistic setting. By doing so they get an excellent physical simulator for realistic small scale testing which can be used to verify a computational simulator. IEEE has created a vehicle-to-vehicle communication standard called IEEE 802.11p: Wireless Access for the Vehicular Environment (WAVE) [Wireless Access in Vehicular Environments]. On the 24th of April 2013, the cross country cooperation Compass4D met in Denmark, where 90 buses will be equipped and tested using this standard for at least one year [http://www.ertico.com/compass4d-project-consortium-meets-in-denmark-where-90-buses-will-be-equipped-and-trialed-for-at-least-one-year/]. These examples are just some of the many great contributions which are aiming towards developing a technology to better the efficiency and safety of roads.In this project, we develop a comprehensive real-time large scale WiFi simulator. It simulates vehicle-to-vehicle and vehicle-to-road communication, and can be a good supplement to systems like the ones in Denmark and Norway. To make large scale testing possible and affordable, we have created a way to generate simplified versions of real life streets and structures using Open Street Maps. The goal of this thesis is to make it possible to simulate dynamic traffic environments with communication in real-time.To achieve this, we harness the compute power of graphics cards which is shown to be extremely powerful in solving massive parallel tasks like ray tracing, the core computational method used in our work. In our case, we use it to trace line-of-sight for the mobile WiFi signals (rather than photon rays). This is done by using the NVIDIA OptiX ray tracing engine for most of the heavy calculations. Using the same framework, we also implement a dynamic environment with both static and moving senders/receivers to illustrate a realistic traffic scenario.Our system has been tested on several benchmarks to examine how it performs in different scenarios. Our results show that it is feasible to created a system capable of simulating medium resolution scenarios with a great number of senders, buildings and moving obstacles in real-time with a frame rate of at least 24fps. We also show that the number of objects, the resolution and even the number of receivers can be increased substantially when simulating vehicle-to- vehicle communication, since it requires lower update rates. Several ideas for how to extend this work is also included.
11
Embley, Ben. "3D loaded foam structures and transport on quasi-1D networks." Thesis, University of Manchester, 2009. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.493534.
Full textAbstract:
This thesis is motivated by a simple idea: the liquid content of a foam, having weight under gravity, can load the structure of bubbles and films that the foam contains. At the simplest level, we can construct a model of a single, loaded channel (called a 'Plateau border') that deforms according to the weight normal to its direction, the liquid content itself resulting from a balance between (tangential) gravity and capillary suction (the 'foam drainage equation'). In this model, only the surface tension of adjoining films supports the applied forces (the 'load').
12
Umar, Abubakar Sadiq. "3D multiple description coding for error resilience over wireless networks." Thesis, Brunel University, 2011. http://bura.brunel.ac.uk/handle/2438/6418.
Full textAbstract:
Mobile communications has gained a growing interest from both customers and service providers alike in the last 1-2 decades. Visual information is used in many application domains such as remote health care, video –on demand, broadcasting, video surveillance etc. In order to enhance the visual effects of digital video content, the depth perception needs to be provided with the actual visual content. 3D video has earned a significant interest from the research community in recent years, due to the tremendous impact it leaves on viewers and its enhancement of the user’s quality of experience (QoE). In the near future, 3D video is likely to be used in most video applications, as it offers a greater sense of immersion and perceptual experience. When 3D video is compressed and transmitted over error prone channels, the associated packet loss leads to visual quality degradation. When a picture is lost or corrupted so severely that the concealment result is not acceptable, the receiver typically pauses video playback and waits for the next INTRA picture to resume decoding. Error propagation caused by employing predictive coding may degrade the video quality severely. There are several ways used to mitigate the effects of such transmission errors. One widely used technique in International Video Coding Standards is error resilience. The motivation behind this research work is that, existing schemes for 2D colour video compression such as MPEG, JPEG and H.263 cannot be applied to 3D video content. 3D video signals contain depth as well as colour information and are bandwidth demanding, as they require the transmission of multiple high-bandwidth 3D video streams. On the other hand, the capacity of wireless channels is limited and wireless links are prone to various types of errors caused by noise, interference, fading, handoff, error burst and network congestion. Given the maximum bit rate budget to represent the 3D scene, optimal bit-rate allocation between texture and depth information rendering distortion/losses should be minimised. To mitigate the effect of these errors on the perceptual 3D video quality, error resilience video coding needs to be investigated further to offer better quality of experience (QoE) to end users. This research work aims at enhancing the error resilience capability of compressed 3D video, when transmitted over mobile channels, using Multiple Description Coding (MDC) in order to improve better user’s quality of experience (QoE). Furthermore, this thesis examines the sensitivity of the human visual system (HVS) when employed to view 3D video scenes. The approach used in this study is to use subjective testing in order to rate people’s perception of 3D video under error free and error prone conditions through the use of a carefully designed bespoke questionnaire.
13
Liau, Forrest (Forrest W. ). "Virus-enabled synthesis and 2D/3D assembly of nanowire networks." Thesis, Massachusetts Institute of Technology, 2013. http://hdl.handle.net/1721.1/92064.
Full textAbstract:
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Materials Science and Engineering, June 2013.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Cataloged from student-submitted PDF version of thesis. "June 2013."
Includes bibliographical references.
Dense networks of high aspect ratio nanowires can provide important functionality to electronic devices through a unique combination of electronic and structural properties including high conductivity, high surface area, and tunable porosity. In this thesis, we explore the virus-enabled synthesis and two/three-dimensional assembly of metallic and semiconducting nanowire networks for future applications including batteries and solar cells. In Chapter 2, we describe the virus-enabled synthesis of titanium oxide nanowires and their incorporation in layer-by-layer polyelectrolyte assemblies for use in dye sensitized solar cells. In Chapter 3, we describe a two-dimensional network of virus-templated cobalt oxide nanowires integrated into ultrathin microbatteries via soft lithography. In Chapter 4, we describe a three-dimensional porous virus-only aerogel network and demonstrate a virus-assembled metal nanowire network for use in batteries. Finally, in Chapter 5, the mechanical properties of various virus assembled three-dimensional structures are measured and compared. We hereby expand the virus assembly toolkit and demonstrate the versatility of bioengineered materials templates.
by Forrest W. Liau.
Ph. D.
14
Prince, Justin. "Coupled 1D-3D simulation of flow in subway transit networks." Thesis, Imperial College London, 2015. http://hdl.handle.net/10044/1/29431.
Full textAbstract:
This thesis presents a 1D-3D coupled approach for tunnel ventilation modelling. A simplified 1D tunnel network model, based on the industry standard tool SES, is developed for testing coupling strategies. For a small tunnel network, air-flow rates are within 2% of those computed using SES. The open-source CFD code OpenFOAM is used to compute pressure losses within the developed coupling approach. Pressure loss calculations using the standard k - ε turbulence model are verified against the commercial code Star-CCM+ and validated against experimental data. In most cases the calculations are within the error bounds of the data. A Lagrange multiplier approach is used to treat defective flow-rate boundary conditions, which arise in multi-dimensional modelling. A novel implementation of this method within the finite-volume framework of OpenFOAM is constructed. The additional unknowns, introduced to the Navier-Stokes equations are solved within an adaption of the PISO solution, used by OpenFOAM. The velocity profiles produced using this method are in excellent agreement with the analytical solutions for laminar pipe flow. Furthermore, a first implementation of the Lagrange multiplier method with turbulence modelling incorporated shows that the approach is stable and accurate for simple and complex flow scenarios. A 1D-3D 'hybrid' coupling strategy has been developed that can be implemented as a 'black box' within the framework of the 1D aerodynamic model, and applicable to SES, by representing the CFD sections of the network as pressure losses. The proposed 1D-3D coupling strategy accurately reproduces full 3D simulations for the same scenario, and also reduces the simulation time. The tools developed in this work allow sophisticated methods to simulate the flow more accurately than 1D tools alone, whilst allowing for a large tunnel network to be modelled at a reasonable cost.
15
Qamar, Isabel Perveen Sophia. "Development of 3D printed vascular networks for repeated self healing." Thesis, University of Bristol, 2017. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.730887.
Full text
16
Christopoulos, Charitos Andreas. "Brain disease classification using multi-channel 3D convolutional neural networks." Thesis, Linköpings universitet, Statistik och maskininlärning, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-174329.
Full textAbstract:
Functional magnetic resonance imaging (fMRI) technology has been used in the investigation of human brain functionality and assist in brain disease diagnosis. While fMRI can be used to model both spatial and temporal brain functionality, the analysis of the fMRI images and the discovery of patterns for certain brain diseases is still a challenging task in medical imaging. Deep learning has been used more and more in medical field in an effort to further improve disease diagnosis due to its effectiveness in discovering high-level features in images. Convolutional neural networks (CNNs) is a class of deep learning algorithm that have been successfully used in medical imaging and extract spatial hierarchical features. The application of CNNs in fMRI and the extraction of brain functional patterns is an open field for research. This project focuses on how fMRIs can be used to improve Autism Spectrum Disorders (ASD) detection and diagnosis with 3D resting-state functional MRI (rs-fMRI) images. ASDs are a range of neurodevelopment brain diseases that mostly affect social function. Some of the symptoms include social and communicating difficulties, and also restricted and repetitive behaviors. The symptoms appear on early childhood and tend to develop in time thus an early diagnosis is required. Finding a proper model for identifying between ASD and healthy subject is a challenging task and involves a lot of hyper-parameter tuning. In this project a grid search approach is followed in the quest of the optimal CNN architecture. Additionally, regularization and augmentation techniques are implemented in an effort to further improve the models performance.
17
Scott, Dion Hamish. "Development of a dynamic 3D imaging system for vascular networks." Thesis, Queensland University of Technology, 2000.
Find full text
18
Vinther, Sven. "Active 3D object recognition using geometric invariants." Thesis, University of Cambridge, 1994. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.362974.
Full text
19
Zhao, Jing. "Protein Structure Prediction Based on Neural Networks." Thèse, Université d'Ottawa / University of Ottawa, 2013. http://hdl.handle.net/10393/23636.
Full textAbstract:
Proteins are the basic building blocks of biological organisms, and are responsible for a variety of functions within them. Proteins are composed of unique amino acid sequences. Some has only one sequence, while others contain several sequences that are combined together. These combined amino acid sequences fold to form a unique three-dimensional (3D) shape. Although the sequences may fold proteins into different 3D shapes in diverse environments, proteins with similar amino acid sequences typically have similar 3D shapes and functions. Knowledge of the 3D shape of a protein is important in both protein function analysis and drug design, for example when assessing the toxicity reduction associated with a given drug. Due to the complexity of protein 3D shapes and the close relationship between shapes and functions, the prediction of protein 3D shapes has become an important topic in bioinformatics.
This research introduces a new approach to predict proteins’ 3D shapes, utilizing a multilayer artificial neural network. Our novel solution allows one to learn and predict the representations of the 3D shape associated with a protein by starting directly from its amino acid sequence descriptors. The input of the artificial neural network is a set of amino acid sequence descriptors we created based on a set of probability density functions. In our algorithm, the probability density functions are calculated by the correlation between the constituent amino acids, according to the substitution matrix. The output layer of the network is formed by 3D shape descriptors provided by an information retrieval system, called CAPRI. This system contains the pose invariant 3D shape descriptors, and retrieves proteins having the closest structures. The network is trained by proteins with known amino acid sequences and 3D shapes. Once the network has been trained, it is able to predict the 3D shape descriptors of the query protein. Based on the predicted 3D shape descriptors, the CAPRI system allows the retrieval of known proteins with 3D shapes closest to the query protein. These retrieved proteins may be verified as to whether they are in the same family as the query protein, since proteins in the same family generally have similar 3D shapes.
The search for similar 3D shapes is done against a database of more than 45,000 known proteins. We present the results when evaluating our approach against a number of protein families of various sizes. Further, we consider a number of different neural network architectures and optimization algorithms. When the neural network is trained with proteins that are from large families where the proteins in the same family have similar amino acid sequences, the accuracy for finding proteins from the same family is 100%. When we employ proteins whose family members have dissimilar amino acid sequences, or those from a small protein family, in which case, neural networks with one hidden layer produce more promising results than networks with two hidden layers, and the performance may be improved by increasing the number of hidden nodes when the networks have one hidden layer.
20
Shafin, Rubayet. "3D Massive MIMO and Artificial Intelligence for Next Generation Wireless Networks." Diss., Virginia Tech, 2020. http://hdl.handle.net/10919/97633.
Full textAbstract:
3-dimensional (3D) massive multiple-input-multiple-output (MIMO)/full dimensional (FD) MIMO and application of artificial intelligence are two main driving forces for next generation wireless systems. This dissertation focuses on aspects of channel estimation and precoding for 3D massive MIMO systems and application of deep reinforcement learning (DRL) for MIMO broadcast beam synthesis. To be specific, downlink (DL) precoding and power allocation strategies are identified for a time-division-duplex (TDD) multi-cell multi-user massive FD-MIMO network. Utilizing channel reciprocity, DL channel state information (CSI) feedback is eliminated and the DL multi-user MIMO precoding is linked to the uplink (UL) direction of arrival (DoA) estimation through estimation of signal parameters via rotational invariance technique (ESPRIT). Assuming non-orthogonal/non-ideal spreading sequences of the UL pilots, the performance of the UL DoA estimation is analytically characterized and the characterized DoA estimation error is incorporated into the corresponding DL precoding and power allocation strategy. Simulation results verify the accuracy of our analytical characterization of the DoA estimation and demonstrate that the introduced multi-user MIMO precoding and power allocation strategy outperforms existing zero-forcing based massive MIMO strategies.
In 3D massive MIMO systems, especially in TDD mode, a base station (BS) relies on the uplink sounding signals from mobile stations to obtain the spatial information for downlink MIMO processing. Accordingly, multi-dimensional parameter estimation of MIMO channel becomes crucial for such systems to realize the predicted capacity gains. In this work, we also study the joint estimation of elevation and azimuth angles as well as the delay parameters for 3D massive MIMO orthogonal frequency division multiplexing (OFDM) systems under a parametric channel modeling. We introduce a matrix-based joint parameter estimation method, and analytically characterize its performance for massive MIMO OFDM systems. Results show that antenna array configuration at the BS plays a critical role in determining the underlying channel estimation performance, and the characterized MSEs match well with the simulated ones. Also, the joint parametric channel estimation outperforms the MMSEbased channel estimation in terms of the correlation between the estimated channel and the real channel.
Beamforming in MIMO systems is one of the key technologies for modern wireless communication. Creating wide common beams are essential for enhancing the coverage of cellular network and for improving the broadcast operation for control signals. However, in order to maximize the coverage, patterns for broadcast beams need to be adapted based on the users' movement over time. In this dissertation, we present a MIMO broadcast beam optimization framework using deep reinforcement learning. Our proposed solution can autonomously and dynamically adapt the MIMO broadcast beam parameters based on user' distribution in the network. Extensive simulation results show that the introduced algorithm can achieve the optimal coverage, and converge to the oracle solution for both single cell and multiple cell environment and for both periodic and Markov mobility patterns.Doctor of Philosophy
Multiple-input-multiple-output (MIMO) is a technology where a transmitter with multiple antennas communicates with one or multipe receivers having multiple antennas. 3- dimensional (3D) massive MIMO is a recently developed technology where a base station (BS) or cell tower with a large number of antennas placed in a two dimensional array communicates with hundreds of user terminals simultaneously. 3D massive MIMO/full dimensional (FD) MIMO and application of artificial intelligence are two main driving forces for next generation wireless systems. This dissertation focuses on aspects of channel estimation and precoding for 3D massive MIMO systems and application of deep reinforcement learning (DRL) for MIMO broadcast beam synthesis. To be specific, downlink (DL) precoding and power allocation strategies are identified for a time-division-duplex (TDD) multi-cell multi-user massive FD-MIMO network. Utilizing channel reciprocity, DL channel state information (CSI) feedback is eliminated and the DL multi-user MIMO precoding is linked to the uplink (UL) direction of arrival (DoA) estimation through estimation of signal parameters via rotational invariance technique (ESPRIT). Assuming non-orthogonal/non-ideal spreading sequences of the UL pilots, the performance of the UL DoA estimation is analytically characterized and the characterized DoA estimation error is incorporated into the corresponding DL precoding and power allocation strategy. Simulation results verify the accuracy of our analytical characterization of the DoA estimation and demonstrate that the introduced multi-user MIMO precoding and power allocation strategy outperforms existing zero-forcing based massive MIMO strategies. In 3D massive MIMO systems, especially in TDD mode, a BS relies on the uplink sounding signals from mobile stations to obtain the spatial information for downlink MIMO processing. Accordingly, multi-dimensional parameter estimation of MIMO channel becomes crucial for such systems to realize the predicted capacity gains. In this work, we also study the joint estimation of elevation and azimuth angles as well as the delay parameters for 3D massive MIMO orthogonal frequency division multiplexing (OFDM) systems under a parametric channel modeling. We introduce a matrix-based joint parameter estimation method, and analytically characterize its performance for massive MIMO OFDM systems. Results show that antenna array configuration at the BS plays a critical role in determining the underlying channel estimation performance, and the characterized MSEs match well with the simulated ones. Also, the joint parametric channel estimation outperforms the MMSE-based channel estimation in terms of the correlation between the estimated channel and the real channel. Beamforming in MIMO systems is one of the key technologies for modern wireless communication. Creating wide common beams are essential for enhancing the coverage of cellular network and for improving the broadcast operation for control signals. However, in order to maximize the coverage, patterns for broadcast beams need to be adapted based on the users' movement over time. In this dissertation, we present a MIMO broadcast beam optimization framework using deep reinforcement learning. Our proposed solution can autonomously and dynamically adapt the MIMO broadcast beam parameters based on user' distribution in the network. Extensive simulation results show that the introduced algorithm can achieve the optimal coverage, and converge to the oracle solution for both single cell and multiple cell environment and for both periodic and Markov mobility patterns.
21
Tannouri, Anthony. "Using Wireless multimedia sensor networks for 3D scene asquisition and reconstruction." Thesis, Bourgogne Franche-Comté, 2018. http://www.theses.fr/2018UBFCD053/document.
Full textAbstract:
De nos jours, les réseaux de capteurs multimédia sans fils sont prometteurs pour différentes applications et domaines, en particulier avec le développement de l’IoT et des capteurs de caméra efficaces et bon marché. La stéréo vision est également très importante pour des objectifs multiples comme la Cinématographie, les jeux, la Réalité Virtuelle, la Réalité Augmentée, etc. Cette thèse vise à développer un système de reconstruction de scène en 3D prouvant l’utilisation de cartes de disparités stéréoscopiques multi-angles dans le contexte des réseaux de capteurs multimedia. Notre travail peut être divisé en trois parties. La première se concentre sur l’étude de toutes les applications, composants, topologies, contraintes et limitations de ces réseaux. En plus, les méthodes de calcul de disparité de vision stéréoscopique afin de choisir la ou les meilleures méthodes pour réaliser une reconstruction en 3D sur le réseau à faible coût en termes de complexité et de consommation d’énergie. Dans la deuxième partie, nous expérimentons et simulons différents calculs de cartes de disparités sur quelques nœuds en changeant les scénarios (intérieur et extérieur), les distances de couverture, les angles, le nombre de nœuds et les algorithmes. Dans la troisième partie, nous proposons un modèle de réseau basé sur l’arbre pour calculer des cartes de disparités précises sur des nœuds de capteurs de caméra multicouches qui répond aux besoins du serveur pour faire une reconstruction de scène 3D de la scène ou de l’objet d’intérêt. Les résultats sont acceptables et assurent la preuve du concept d’utilisation des cartes de disparités dans le contexte des réseaux de capteurs multimédiaNowadays, the WMSNs are promising for different applications and fields, specially with the development of the IoT and cheap efficient camera sensors. The stereo vision is also very important for multiple purposes like Cinematography, games, Virtual Reality, Augmented Reality, etc. This thesis aim to develop a 3D scene reconstruction system that proves the concept of using multiple view stereo disparity maps in the context of WMSNs. Our work can be divided in three parts. The first one concentrates on studying all WMSNs applications, components, topologies, constraints and limitations. Adding to this stereo vision disparity map calculations methods in order to choose the best method(s) to make a 3d reconstruction on WMSNs with low cost in terms of complexity and power consumption. In the second part, we experiment and simulate different disparity map calculations on a couple of nodes by changing scenarios (indoor and outdoor), coverage distances, angles, number of nodes and algorithms. In the third part, we propose a tree-based network model to compute accurate disparity maps on multi-layer camera sensor nodes that meets the server needs to make a 3d scene reconstruction of the scene or object of interest. The results are acceptable and ensure the proof of the concept to use disparity maps in the context of WMSNs
22
Neupane, Bishal, and Pooya Moazzeni. "Live 3D-TV Streaming." Thesis, Blekinge Tekniska Högskola, Sektionen för ingenjörsvetenskap, 2012. http://urn.kb.se/resolve?urn=urn:nbn:se:bth-4016.
Full textAbstract:
The world is not flat as a pancake. It has height, width and depth. So we should see it even on TV. So far we cannot see three-dimensional programs directly into our TVs. Not even in cinemas with 3D cinema works "for real". For still there the magic sits in those glasses. The glasses of different colors allow distinguishing right and left eye impression tightened so that one sees different images with each eye. That is what creates the illusion of three dimensions. The goal of this thesis is to be on track to change that. Then you should achieve the same feeling without having required glasses, however, with a different technique. Do you remember those pictures that used to accompany the cereal packets? When angled in one direction, it was Donald Duck and angled it the other way it was Mickey Mouse. Our work is in the same way, though not with different images but with different perspectives. Same ribbed surface that existed at the pictures in cereal packets, are used as matter of fact on our 3D TV screen. Depending from which angle you look certain image information is hiding as it falls behind the ribbed surface. It thus separates views through the screen. This thesis project is focused on a prototyping of live 3D TV streaming application where a live video of a scene is viewed on a 3D auto-stereoscopic display that gives two different perspectives, or views, simultaneously. The TV uses a face search (eye tracking) system to set up the television optimal for those who want to see 3D without glasses. During thesis a simple 3D studio was built where the focus has been to show depth perception. For scene capturing two cameras were used. We have found an engineering solution to take pictures simultaneously from the cameras. The input images from two cameras are sent to an analog to digital converter (frame grabber) as two channels of a virtual color camera, which means real time and synchronized capturing in a simple way. The project has several applications written in C++ using various open source libraries, which essentially grab stereo image sequences from cameras using frame grabber, transfer image sequences to other applications via server communication, and display the live video in 3D display by exclusive rendering method. The communications between different applications for the purposes of transmission and receiving of video data is done using socket programming. The results of the project are very promising in which the live video of a scene can be viewed with noticeable depth despite obvious lagging in video timing.
23
Appuhami, Ralalage Harsha Nishantha Deepal. "Cross-layer design for scalable/3D wireless video transmission over IP networks." Thesis, Kingston University, 2014. http://eprints.kingston.ac.uk/30012/.
Full textAbstract:
The first two parts of the thesis address the issues related to 3D video transmission over wireless networks and proposes cross-layer design techniques to optimise the information exchange between dif- ferent Open Systems Interconnection (OSI) layers or system blocks. In particular, the first section of this thesis exploits the flexibility of adjusting the checksum coverage length of the transport layer pro- tocol, UDP-lite as opposed to its counterpart UDP. The study pro- poses an optimum checksum coverage length to protect only impor- tant header information of an H.264 encoded video transmission over wireless links, together with robust header compression (RoHC) and Automatic Retransmission Request (ARQ). The second part of the thesis investigates a content and Channel aware Medium Access Con- trol (MAC) layer scheduling algorithm by considering the layer prior- ities of an H.264 Scalable Video Coding (SVC) encoded 3D video transmission over an Orthogonal Frequency Division Multiple Ac- cess (OFDMA) based wireless link with a prioritised queuing tech- nique to improve the Quality of Experience (QoE) of the end users. A considerable amount of research time was devoted to investigat- ing accurate, consistent and real-time quality evaluation techniques for 3D image/ video as cross-layer design techniques mostly rely on the quality feedbacks from end users to optimise system parameters. The first quality metric proposed is a stereoscopic image quality met- ric using the disparity histogram of the left and right views. A 3D stereoscopic video quality evaluation technique is proposed, based on the predominant energy distribution of gradients using 3D structural tensors in the next section. Finally, a near no reference quality metric is proposed for colour plus depth 3D video compression and transmis- sion, using the extracted edge information of colour images and depth maps. The research investigates a number of error resilient transmission methods to combat artifacts in 3D video delivery over wireless chan- nels. A Region-of-Interest (ROI) based transmission method for stereo- scopic videos has been proposed to mark the important areas of the video and provide Unequal Error Protection (UEP) during transmis- sion. Next, we investigate the effects of compression and packet loss on the rendered video quality and propose a model to quantify ren- dering and concealment errors at the sender-side and then use the information generated through the model to effectively deliver 3D. Finally an asymmetric coding approach is suggested for 3D medical video transmitted over band limited wireless networks by considering large data rates associated with 3D medical video as they are usually captured in high resolution and pixel depth. Key words: 3D video transmission, Cross-layer design, Orthogonal frequency-division multiple access, H.264 video compression, Scalable video coding, Robust header compression, automatic retransmission request, Quality of experience, Prioritized 3D video transmission, Un- equal error protection.
24
Sällqvist, Jessica. "Real-time 3D Semantic Segmentation of Timber Loads with Convolutional Neural Networks." Thesis, Linköpings universitet, Datorseende, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-148862.
Full textAbstract:
Volume measurements of timber loads is done in conjunction with timber trade. When dealing with goods of major economic values such as these, it is important to achieve an impartial and fair assessment when determining price-based volumes. With the help of Saab’s missile targeting technology, CIND AB develops products for digital volume measurement of timber loads. Currently there is a system in operation that automatically reconstructs timber trucks in motion to create measurable images of them. Future iterations of the system is expected to fully automate the scaling by generating a volumetric representation of the timber and calculate its external gross volume. The first challenge towards this development is to separate the timber load from the truck. This thesis aims to evaluate and implement appropriate method for semantic pixel-wise segmentation of timber loads in real time. Image segmentation is a classic but difficult problem in computer vision. To achieve greater robustness, it is therefore important to carefully study and make use of the conditions given by the existing system. Variations in timber type, truck type and packing together create unique combinations that the system must be able to handle. The system must work around the clock in different weather conditions while maintaining high precision and performance.
25
Wiklander, Marcus. "Classification of tree species from 3D point clouds using convolutional neural networks." Thesis, Umeå universitet, Institutionen för fysik, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-174662.
Full textAbstract:
In forest management, knowledge about a forest's distribution of tree species is key. Being able to automate tree species classification for large forest areas is of great interest, since it is tedious and costly labour doing it manually. In this project, the aim was to investigate the efficiency of classifying individual tree species (pine, spruce and deciduous forest) from 3D point clouds acquired by airborne laser scanning (ALS), using convolutional neural networks. Raw data consisted of 3D point clouds and photographic images of forests in northern Sweden, collected from a helicopter flying at low altitudes. The point cloud of each individual tree was connected to its representation in the photos, which allowed for manual labeling of training data to be used for training of convolutional neural networks. The training data consisted of labels and 2D projections created from the point clouds, represented as images. Two different convolutional neural networks were trained and tested; an adaptation of the LeNet architecture and the ResNet architecture. Both networks reached an accuracy close to 98 %, the LeNet adaptation having a slightly lower loss score for both validation and test data compared to that of ResNet. Confusion matrices for both networks showed similar F1 scores for all tree species, between 97 % and 98 %. The accuracies computed for both networks were found higher than those achieved in similar studies using ALS data to classify individual tree species. However, the results in this project were never tested against a true population sample to confirm the accuracy. To conclude, the use of convolutional neural networks is indeed an efficient method for classification of tree species, but further studies on unbiased data is needed to validate these results.
26
Dogru, Sedat. "Sycophant Wireless Sensor Networks Tracked By Sparsemobile Wireless Sensor Networks While Cooperativelymapping An Area." Phd thesis, METU, 2012. http://etd.lib.metu.edu.tr/upload/12615139/index.pdf.
Full textAbstract:
In this thesis the novel concept of Sycophant Wireless Sensors (SWS) is introduced. A SWS network is a static ectoparasitic clandestine sensor network mounted incognito on a mobile agent using only the agent&rsquos mobility without intervention. SWS networks not only communicate with each other through mobileWireless Sensor Networks (WSN) but also cooperate with them to form a global hybrid Wireless Sensor Network. Such a hybrid network has its own problems and opportunities, some of which have been studied in this thesis work. Assuming that direct position measurements are not always feasible tracking performance of the sycophant using range only measurements for various communication intervals is studied. Then this framework was used to create a hybrid 2D map of the environment utilizing the capabilities of the mobile network the sycophant. In order to show possible applications of a sycophant deployment, the sycophant sensor node was equipped with a laser ranger as its sensor, and it was let to create a 2D map of its environment. This 2D map, which corresponds to a height dierent than the follower network, was merged with the 2D map of the mobile network forming a novel rough 3D map. Then by giving up from the need to properly localize the sycophant even when it is disconnected to the rest of the network, a full 3D map of the environment is obtained by fusing 2D map and tracking capabilities of the mobile network with the 2D vertical scans of the environment by the sycophant. And finally connectivity problems that arise from the hybrid sensor/actuator network were solved. For this 2 new connectivity maintenance algorithms, one based on the helix structures of the proteins, and the other based on the acute triangulation of the space forming a Gabriel Graph, were introduced. In this new algorithms emphasis has been given to sparseness in order to increase fault tolerance to regional problems. To better asses sparseness a new measure, called Resistance was introduced, as well as another called updistance.
27
Rowe, Laura Elizabeth. "An Active Microscaffold System with Fluid Delivery and Stimulation/Recording Functionalities for Culturing 3-D Neuronal Networks." Diss., Georgia Institute of Technology, 2007. http://hdl.handle.net/1853/14542.
Full textAbstract:
An Active Microscaffold System with Fluid Delivery and Stimulation/Recording Functionalities for Culturing 3-D Neuronal Networks
Laura Elizabeth Rowe
215 Pages
Directed by Dr. A. Bruno Frazier
An active microscaffold system with fluid delivery and electrical stimulation/recording functionalities for 3-D neuronal culture studies is presented. The microscaffolds presented in this dissertation consist of an array of microfabricated towers with integrated microfluidic channels, fluid ports, and electrodes. The microfluidic channels and ports allow for perfusion of nutrients, gas exchange, and biochemical control of the extracellular environment throughout the 3-D culture, while the electrodes allow for active stimulation/recording of the 3-D neuronal network. In essence, the microscaffold serves as an artificial circulatory system to enable 3-D in vitro growth and proliferation of re-aggregate neuronal cell cultures. Increased cell survival on microscaffolds with nutrient perfusion at 14 and 21 days in vitro (DIV) is presented. Additionally, the microtower scaffold is built upon a substrate that is compatible with the Multi Channel Systems preamplifier setup to enable electrical stimulation/recording of the cultured network in a 3-D mutilelectrode array (MEA) environment. Impedance measurements on the functioning microtower electrodes were obtained. The overall goal of this research was to develop a BioMEMS technology to provide neuroscientists with a better investigative tool for studying 3-D in vitro neuronal networks than is currently available.
28
Justin, Alexander William. "Vascular network formation via 3D printing and cell-based approaches." Thesis, University of Cambridge, 2018. https://www.repository.cam.ac.uk/handle/1810/276227.
Full textAbstract:
Vascularization is essential for living tissue and remains a major challenge in the field of tissue engineering. A lack of a perfusable channel network within a large and densely populated tissue engineered construct leads to necrotic core formation, preventing fabrication of functional tissues and organs. While many approaches have been reported for forming vascular networks, including materials processing techniques, such those involving lithography, bioprinting, and sacrificial templating; and cell-based approaches, in which cellular self-organization processes form vessels; all are deficient in their ability to form a vessel system of sufficient complexity for supporting a large cellular construct. What is missing from the literature is a method for forming a fully three-dimensional vascular network over the full range of length-scales found in native vessel systems, which can be used alongside cells and perfused with fluids to support their function. A large number of research groups are thus pursuing novel methods for fabricating vascular systems in order that new tissues and organs can be fabricated in the lab. In this project, a 3D printing-based approach was used to form vascular networks which are hierarchical, three-dimensional, and perfusable. This was performed in thick, cellularized hydrogels similar in composition to native tissue; these being collagen (ECM-like) and fibrin (woundlike), both of which are highly capable of supporting cellular activities, such as cell seeding, cell spreading, and capillary morphogenesis. In order to make use of 3D printed network templates in cellularized hydrogel environments, it was necessary to develop a new approach in which standard 3D printed materials were converted into a gelatin template, via an alginate intermediary, which can be removed quickly in physiologic conditions and which does not reduce cell viability. This multi-casting approach enables a hierarchical channel network to be formed in three-dimensions, capable of being perfused with cell medium to maintain the viability of a cell population, thereby addressing the fundamental problem. Using standard cell staining and immuno-histochemistry techniques, we showed good endothelial cell seeding and the presence of tight junctions between the channel endothelial cells. When fibroblasts were seeded into the bulk of the hydrogel, a high degree of cell viability and cell spreading was observed when a threshold flow rate is met. By counting the number of live and dead cells in a sample regions of the gel, we were able to show a dependency of cell viability upon the perfusion flow rate and further determine a regime in which the vast majority of cells are alive and spreading. This data informs future cellular experiments using this platform technology. The limits of existing 3D printing technology meant that the micro-scale vasculature needed to be formed by other means. Cellular co-culture of endothelial and stromal cell types has been shown to be capable of forming capillary-like structures in vitro. For inclusion with the 3D printed channel system, we investigated the use of an angiogenic method for capillary formation, using multi-cellular spheroids, and a vasculogenic approach, using individual cells, in order that the full vascular system could be constructed. Endothelial and mesenchymal stromal cells were encapsulated in small fibrin and collagen gels and maintained under static culture conditions in order to form capillaries by the above approaches. The aim here was to find a particular gel composition and cell concentration which would support capillary morphogenesis while being suitably robust to handle the mechanical stresses associated with perfusion. As future work, the next step will be to incorporate the vasculogenic co-culture technique, used to form capillary-sized vessels, into a perfusable gel containing the large templated channels, formed via the multi-casting approach. The challenge here is to anastomose the capillary-sized vessels to the large templated channels and thereby enable perfusion of the capillary vessels. This step would be a highly significant development in the field as it would mean large constructs could be fabricated with physiological densities of cells, which could lead to a range of potential therapeutic applications.
29
Maamar, Haifa Raja. "Supplying Partners Suite of Protocols for P2P 3D Streaming Over Thin Mobile Devices." Thèse, Université d'Ottawa / University of Ottawa, 2013. http://hdl.handle.net/10393/23709.
Full textAbstract:
The recent advances in mobile computing devices and wireless networking produced the technical platform for multimedia services over thin mobile devices. Nowadays, we are witnessing an important growth in applications using thin mobile devices, such as social networks, virtual walkthrough, media streaming, and augmented reality (AR), just to mention a few. Most of these applications are based on the client-server architecture, however several studies showed that the client-server architecture suffers from various issues, such as the server bottleneck, latency and the lack of scalability. This led most of the systems to switch to the peer-to-peer (P2P)-like environment for its scalability and potential cost saving.
P2P multimedia streaming over thin mobile devices-based classes of applications has known a significant growth during the last years. Although P2P video streaming over thin mobile devices received a great deal of attention, the application of 3D streaming over mobile devices was challenging mainly due to the limited mobile resources and capabilities, as well as the wireless medium limitations. Having 3D streaming over Mobile Ad hoc Networks (MANET) is considered more challenging given that the 3D streaming-based system has to deal with a dynamic environment resulting from nodes mobility, which may lead to route breakages and connection loss. Therefore, one of the major difficulties in 3D streaming over MANET is related to the supplying partner's strategy that aims at determining the most suitable source holding the required 3D data to stream it quickly and efficiently to the requesters. In this thesis, we propose our P2P based 3D streaming system which we refer to as MOSAIC as well as a suite of supplying partner strategy protocols for P2P 3D streaming over thin mobile devices. Our proposed suite of protocols selects the potential sources that have the relevant 3D data, based on a set of criteria such as the source location, the mobile device's available resources as well as its residual energy. We also proposed a multihop supplying partner selection protocol that takes into account the signal strength and the nodes mobility when streaming the relevant 3D data. The performance evaluation obtained to evaluate our MOSAIC system as well as our suite of protocols using an extensive set of NS2 simulation experiments, is then reported.
30
Andriolo, Stefano. "Convolutional Neural Networks in Tomographic Image Enhancement." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/22843/.
Full textAbstract:
Convolutional Neural Networks have seen a huge rise in popularity in image applications. They have been used in medical imaging contexts to enhance the overall quality of the digital representation of the patient's scanned body region and have been very useful when dealing with limited-angle tomographic data. In this thesis, a particular type of convolutional neural network called Unet will be used as the starting point to explore the effectiveness of different networks in enhancing tomographic image reconstructions. We will first make minor tweaks to the 2-dimensional convolutional network and train it on two different datasets. After that, we will take advantage of the shape of the reconstructions we are considering to extend the convolutions to the third dimension. The scanner layout that has been considered for projecting and reconstructing volumes in this thesis indeed consits of a cone-beam geometry, whose output is a volume that approximates the original scanned object. We will then discuss the results in order to try to understand if the proposed solutions could be viable approaches for enhancing tomographic images.
31
Groueix, Thibault. "Learning 3D Generation and Matching." Thesis, Paris Est, 2020. http://www.theses.fr/2020PESC1024.
Full textAbstract:
L'objectif de cette thèse est de développer des approches d'apprentissage profond pour modéliser et analyser les formes 3D. Les progrès dans ce domaine pourraient démocratiser la création artistique d'actifs 3D, actuellement coûteuse en temps et réservés aux experts du domaine. Nous nous concentrons en particulier sur deux tâches clefs pour la modélisation 3D : la reconstruction à vue unique et la mise en correspondance de formes.Une méthode de reconstruction à vue unique (SVR) prend comme entrée une seule image et prédit le monde physique qui a produit cette image. SVR remonte aux premiers jours de la vision par ordinateur. Étant donné que plusieurs configurations de formes, de textures et d'éclairage peuvent expliquer la même image il faut formuler des hypothèses sur la distribution d'images et de formes 3D pour résoudre l’ambiguïté. Dans cette thèse, nous apprenons ces hypothèses à partir de jeux de données à grande échelle au lieu de les concevoir manuellement. Les méthodes d'apprentissage nous permettent d'effectuer une reconstruction complète et réaliste de l'objet, y compris des parties qui ne sont pas visibles dans l'image d'entrée.La mise en correspondance de forme vise à établir des correspondances entre des objets 3D. Résoudre cette tâche nécessite à la fois une compréhension locale et globale des formes 3D qui est difficile à obtenir explicitement. Au lieu de cela, nous entraînons des réseaux neuronaux sur de grands jeux de données pour capturer ces connaissances implicitement.La mise en correspondance de forme a de nombreuses applications en modélisation 3D telles que le transfert d'attribut, le gréement automatique pour l'animation ou l'édition de maillage.La première contribution technique de cette thèse est une nouvelle représentation paramétrique des surfaces 3D modélisées par les réseaux neuronaux. Le choix de la représentation des données est un aspect critique de tout algorithme de reconstruction 3D. Jusqu'à récemment, la plupart des approches profondes en génération 3D prédisaient des grilles volumétriques de voxel ou des nuages de points, qui sont des représentations discrètes. Au lieu de cela, nous présentons une approche qui prédit une déformation paramétrique de surface, c'est-à-dire une déformation d'un modèle source vers une forme objectif. Pour démontrer les avantages ses avantages, nous utilisons notre nouvelle représentation pour la reconstruction à vue unique. Notre approche, baptisée AtlasNet, est la première approche profonde de reconstruction à vue unique capable de reconstruire des maillages à partir d'images sans s’appuyer sur un post-traitement indépendant, et peut le faire à une résolution arbitraire sans problèmes de mémoire. Une analyse plus détaillée d’AtlasNet révèle qu'il généralise également mieux que les autres approches aux catégories sur lesquelles il n'a pas été entraîné.Notre deuxième contribution est une nouvelle approche de correspondance de forme purement basée sur la reconstruction par des déformations. Nous montrons que la qualité des reconstructions de forme est essentielle pour obtenir de bonnes correspondances, et donc introduisons une optimisation au moment de l'inférence pour affiner les déformations apprises. Pour les humains et d'autres catégories de formes déformables déviant par une quasi-isométrie, notre approche peut tirer parti d'un modèle et d'une régularisation isométrique des déformations. Comme les catégories présentant des variations non isométriques, telles que les chaises, n'ont pas de modèle clair, nous apprenons à déformer n'importe quelle forme en n'importe quelle autre et tirons parti des contraintes de cohérence du cycle pour apprendre des correspondances qui respectent la sémantique des objets. Notre approche de correspondance de forme fonctionne directement sur les nuages de points, est robuste à de nombreux types de perturbations, et surpasse l'état de l'art de 15% sur des scans d'humains réelsThe goal of this thesis is to develop deep learning approaches to model and analyse 3D shapes. Progress in this field could democratize artistic creation of 3D assets which currently requires time and expert skills with technical software.We focus on the design of deep learning solutions for two particular tasks, key to many 3D modeling applications: single-view reconstruction and shape matching.A single-view reconstruction (SVR) method takes as input a single image and predicts the physical world which produced that image. SVR dates back to the early days of computer vision. In particular, in the 1960s, Lawrence G. Roberts proposed to align simple 3D primitives to the input image under the assumption that the physical world is made of cuboids. Another approach proposed by Berthold Horn in the 1970s is to decompose the input image in intrinsic images and use those to predict the depth of every input pixel.Since several configurations of shapes, texture and illumination can explain the same image, both approaches need to form assumptions on the distribution of images and 3D shapes to resolve the ambiguity. In this thesis, we learn these assumptions from large-scale datasets instead of manually designing them. Learning allows us to perform complete object reconstruction, including parts which are not visible in the input image.Shape matching aims at finding correspondences between 3D objects. Solving this task requires both a local and global understanding of 3D shapes which is hard to achieve explicitly. Instead we train neural networks on large-scale datasets to solve this task and capture this knowledge implicitly through their internal parameters.Shape matching supports many 3D modeling applications such as attribute transfer, automatic rigging for animation, or mesh editing.The first technical contribution of this thesis is a new parametric representation of 3D surfaces modeled by neural networks.The choice of data representation is a critical aspect of any 3D reconstruction algorithm. Until recently, most of the approaches in deep 3D model generation were predicting volumetric voxel grids or point clouds, which are discrete representations. Instead, we present an alternative approach that predicts a parametric surface deformation ie a mapping from a template to a target geometry. To demonstrate the benefits of such a representation, we train a deep encoder-decoder for single-view reconstruction using our new representation. Our approach, dubbed AtlasNet, is the first deep single-view reconstruction approach able to reconstruct meshes from images without relying on an independent post-processing, and can do it at arbitrary resolution without memory issues. A more detailed analysis of AtlasNet reveals it also generalizes better to categories it has not been trained on than other deep 3D generation approaches.Our second main contribution is a novel shape matching approach purely based on reconstruction via deformations. We show that the quality of the shape reconstructions is critical to obtain good correspondences, and therefore introduce a test-time optimization scheme to refine the learned deformations. For humans and other deformable shape categories deviating by a near-isometry, our approach can leverage a shape template and isometric regularization of the surface deformations. As category exhibiting non-isometric variations, such as chairs, do not have a clear template, we learn how to deform any shape into any other and leverage cycle-consistency constraints to learn meaningful correspondences. Our reconstruction-for-matching strategy operates directly on point clouds, is robust to many types of perturbations, and outperforms the state of the art by 15% on dense matching of real human scans
32
Gustafsson, Klas, and Oskar Berg. "2D and 3D Visualization to Support Fieldwork in the Area of Utility Networks." Thesis, KTH, Geoinformatik, 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-221430.
Full textAbstract:
Utility network fieldworkers of today want to access more information and can benefit a lot from new technical development. Today most fieldwork is conducted using paper plans or locally stored data on laptops as a visual aid. Therefore there is a need for improvement and development of new reliable software for fieldwork. Also the abil- ity to use advanced Geographic Information Systems (GIS) solutions and enhanced visualization methods while out in the field could help improve fieldwork. In order to be as e↵ective as possible when carrying out di↵erent tasks in the field, di↵erent ways of visualizing the same network data are required. 2D and 3D visualization methods have di↵erent advantages and disadvantages when it comes to visualizing network data, which will be accounted for in this thesis. There are three main objectives in this thesis. The first is to evaluate how suitable di↵erent visualization methods are for fieldwork users working with utility networks. The second is to get a better understanding of what hardware and software that can be used for implementing the visualization methods. The last one is to use the first and second objectives to develop a prototype for utility network fieldwork. To address the objectives, the first step is to understand the users that work in the field. By conducting interviews, information about the current workflow for fieldworkers and their opinions about how the systems currently work is gathered. Based on this information the thesis is divided into cases and criteria which is the foundation for proposing a solution in form of mock-up sketches which is then imple- mented in form of a prototype. Finally the prototype is evaluated quantitatively and qualitatively using a web survey and presentations for potential end users. The prototype is created using web technologies and is mainly intended for tablets. Because of its mobility, screen size and adequate computational power the tablet is a good hardware choice for conducting fieldwork. The prototype presents network data in a 2D interactive map view, a 3D augmented reality (AR) view and a combined view. These choices are based on information gathered by studying related work and performing interviews with potential end users in the beginning of the study. The results of the thesis highlights large possibilities in making field work more e↵ective for fieldworkers. This in concluded partly by the results of the interviews with potential end users, but also by the response of the survey and presentation of the suggested solution. It is shown that there are new ways to improve the work process out in the field and that AR can help in visualizing the network in a new informative way for fieldwork. However, several challenges remain, but rapid techno- logical development implies possible solutions to deal with these challenges.
33
Ziqing, Li S. "Towards 3D vision from range images : an optimisation framework and parallel distributed networks." Thesis, University of Surrey, 1991. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.291880.
Full text
34
Teratani, Mehrdad. "3D Image Processing and Communication in Camera Sensor Networks: Free Viewpoint Television Networking." Doctoral thesis, Nagoya University, Nagoya, Japan, 2004. http://hdl.handle.net/2013/ULB-DIPOT:oai:dipot.ulb.ac.be:2013/319673.
Full text
35
Grundberg, Måns, and Viktor Altintas. "Generating 3D Scenes From Single RGB Images in Real-Time Using Neural Networks." Thesis, Malmö universitet, Institutionen för datavetenskap och medieteknik (DVMT), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-43091.
Full textAbstract:
The ability to reconstruct 3D scenes of environments is of great interest in a number of fields such as autonomous driving, surveillance, and virtual reality. However, traditional methods often rely on multiple cameras or sensor-based depth measurements to accurately reconstruct 3D scenes. In this thesis we propose an alternative, deep learning-based approach to 3D scene reconstruction for objects of interest, using nothing but single RGB images. We evaluate our approach using the Deep Object Pose Estimation (DOPE) neural network for object detection and pose estimation, and the NVIDIA Deep learning Dataset Synthesizer for synthetic data generation. Using two unique objects, our results indicate that it is possible to reconstruct 3D scenes from single RGB images within a few centimeters of error margin.
36
Karlsson, Linda Sofia. "Coding for Improved Perceived Quality of 2D and 3D Video over Heterogeneous Networks." Doctoral thesis, Mittuniversitetet, Institutionen för informationsteknologi och medier, 2010. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-11461.
Full textAbstract:
The rapid development of video applications for TV, the internet and mobile phones is being taken one step further in 2010 with the introduction of stereo 3D TV. The 3D experience can be further improved using multiple views in the visualization. The transmission of 2D and 3D video at a sufficiently perceived quality is a challenge considering the diversity in content, the resources of the network and the end-users.Two problems are addressed in this thesis. Firstly, how to improve the perceived quality for an application with a limited bit rate. Secondly, how to ensure the best perceived quality for all end-users in a heterogeneous network. A solution to the first problem is region-of-interest (ROI) video coding, which adapts the coding to provide a better quality in regions of interest to the viewer. A spatio-temporal filter is proposed to provide codec and standard independent ROI video coding. The filter reduces the number of bits necessary to encode the background and successfully re-allocate these bits to the ROI. The temporal part of the filter reduces the complexity compared to only using a spatial filter. Adaption to the requirements of the transmission channel is possible by controlling the standard deviation of the filter. The filter has also been successfully applied to 3D video in the form of 2D-plus-depth, where the depth data was used in the detection of the ROI. The second problem can be solved by providing a video sequence that has the best overall quality. Hence, the best quality for each part of the network and for each 2D and 3D visualization system over time. Scalable video coding enables the extraction of the parts of the data to adapt to the requirements of the network and the end-user. A scheme is proposed in this thesis that provides scalability in the depth and view domain of multi-view plus depth video. The data are divided into enhancement layers depending on the content’s distance to the camera. Schemes to divide the data into layers within a view and between adjacent views have been analyzed. The quality evaluation indicates that the position of the layers in depth as well as the number of layers should be determined by analyzing the depth distribution. The front-most layers in adjacent views should be given priority over the others unless the application requires a high quality of the center views.Den snabba utvecklingen av videoapplikation för TV, Internet och mobiltelefoner tar ytterliggare ett steg i och med introduceringen av stereo 3D TV under 2010. Upplevelsen av 3D kan förstärkas ytterliggare genom att använda multipla vyer i visualiseringen. Skillnaden i innehåll, nätverksresurser och slutanvändare gör överföring av 2D och 3D video med en tillräcklig hög upplevd kvalitet till en utmaning. För det första, hur man ökar den upplevda kvalitén hos en applikation med en begränsad överföringshastighet. För det andra, hur man tillhandahåller den bästa upplevda kvalitén hos alla slutanvändare i ett heterogent nätverk. Region-of-interest (ROI) videokodning är en lösning till det första problemet, vilken anpassar kodningen för att ge högre kvalitet i regioner som är intressanta för användaren. Ett spatio-temporalt filter är föreslaget för att tillhandahålla codec- och standardoberoende ROI videokodning. Filtret reducerar antalet bitar som krävs för att koda bakgrunden och omfördelar dessa till ROI:t. Den temporala delen av filtret minskar komplexiteten jämfört med att använda enbart spatiala filter. Filtret kan anpassas till överföringshastigheten genom att ändra standardavvikelsen för filtret. Filtret har också˙ använts på˙ 3D video i formen 2D-plus-depth, där djupdata användes i ROI detektionen. Det andra problemet kan lösas genom att tillhandahålla en videosekvens som har högsta möjliga kvalitet i hela nätverket. Därmed även den bästa kvaliteten för for varje del av nätverket och för varje 2D- och 3D-skärm. Skalbar videokodning gör det möjligt att extrahera delar av datan för anpassning till de rådande förutsättningarna. En metod som ger skalbarhet i djupet och mellan kameravyer hos multi-view plus depth video har föreslagits. Videosekvensen delas upp i lager beroende på innehållets avstånd till kameran. Metoder för att fördela data över lager i djupet och mellan närliggande vyer har analyserats. Kvalitetsutvärderingen visar att lagrens position i djupet och antalet lager bör bestämmas utifrån fördelningen av djupdata. De främsta lagren i närliggande vyer bör ges högre prioritet om inte applikationen kräver hög kvalitet hos vyer i centrum.
Medi3D
3D-reklam
MediaSense
37
MORENA, Anthony. "POSS-based 3D functional networks as catalysts for the conversion of carbon dioxide." Doctoral thesis, Università degli Studi di Palermo, 2023. https://hdl.handle.net/10447/580035.
Full textAbstract:
Oggi lo sviluppo di processi sostenibili è al centro dell'attenzione a causa delle emergenze climatiche. La Chimica Verde, con i suoi dodici principi sviluppati da Paul Anastas, si concentra sullo sviluppo di processi alternativi e più rispettosi dell'ambiente. Questa branca della chimica mostra come concetti quali la prevenzione dei rifiuti, l'uso di materie prime rinnovabili e la catalisi siano di grande importanza per rendere un processo più sostenibile. In questo contesto, l'anidride carbonica (CO2) rappresenta una delle materie prime più abbondanti, non tossiche e rinnovabili. La possibilità di riutilizzare la CO2 e di trasformare questa molecola in prodotti a valore aggiunto come i carbonati ciclici, attraverso l'addizione in epossidi, rende questa molecola interessante dal punto di vista della ricerca. Nonostante queste interessanti caratteristiche, la trasformazione della CO2 ha un requisito termodinamico energetico molto elevato. Per superare questo problema, è essenziale progettare un catalizzatore adatto per diminuire l'energia di attivazione del processo e renderlo eseguibile in condizioni più blande. In questo contesto si inserisce la presente tesi di dottorato. Tra tutti i possibili catalizzatori proposti in letteratura, una classe di catalizzatori poco esplorata è quella dei materiali bifunzionali. Le eccellenti prestazioni catalitiche mostrate da questi catalizzatori eterogenei sembrano ridurre il divario in termini di attività catalitica tra catalisi eterogenea e omogenea in relazione a questa reazione.
Il primo capitolo della tesi introduce il lettore nel campo della catalisi eterogenea e nella catalisi applicata alla conversione dell'anidride carbonica in carbonati ciclici. Dopo questo capitolo introduttivo, la tesi si divide in due sezioni principali.
Una prima parte, il Capitolo III, riguarda lo studio della stabilità durante la catalisi di diversi materiali basati su nanoforme di carbonio (CNF). D'altra parte, i capitoli IV e V sono relativi alla seconda parte della tesi che pone la sua attenzione sulla progettazione di diversi materiali bifunzionali basati sui silsesquiossani poliedrici oligomerici e sulla loro applicazione come catalizzatori per la conversione di anidride carbonica con epossidi.Today, the development of sustainable processes is in the spotlight due to the climate emergencies. Green Chemistry with its twelve principles developed by Paul Anastas, focuses on the development of alternative and more environmentally friendly processes. This branch of chemistry shows how concepts such as waste prevention, the use of renewable raw materials and catalysis are of great importance in making a process more sustainable. In this context, carbon dioxide (CO2) represents one of the most abundant, non-toxic and renewable carbon feedstocks. The possibility of reusing CO2 and transforming this molecule into value-added products such as cyclic carbonates, via addition into epoxides, makes this molecule interesting from a research point of view. Despite these interesting characteristics, the transformation of CO2 has a very high energy thermodynamic requirement. To overcome this problem, a suitable catalyst is essential to decrease the activation energy of the process and make it work in milder condition. In this context this Ph.D. dissertation finds its outlines. Among all the possible catalysts proposed in literature, one class of catalysts that has been little explored is that of bifunctional materials. The excellent catalytic performance shown by these heterogeneous catalysts seems to narrow the gap in terms of catalytic activity between heterogeneous and homogeneous catalysis in relation to this reaction. The first Chapter of the dissertation introduces the reader into the field of heterogeneous catalysis and in catalysis applied for the conversion of carbon dioxide in cyclic carbonates. After this introductory chapter, the dissertation is splitted into two main sections. A first part concerning the Chapter III is related to the study of the stability during catalysis of different materials based on carbon nanoforms (CNFs). On the other hand, Chapters IV and V are related to the second part of the thesis that put its attention in the design of different bifunctional materials based on polyhedral oligomeric silsesquioxanes and their application as a catalyst for the conversion of carbon dioxide with epoxides.
38
Chen, Cong. "High-Dimensional Generative Models for 3D Perception." Diss., Virginia Tech, 2021. http://hdl.handle.net/10919/103948.
Full textAbstract:
Modern robotics and automation systems require high-level reasoning capability in representing, identifying, and interpreting the three-dimensional data of the real world. Understanding the world's geometric structure by visual data is known as 3D perception. The necessity of analyzing irregular and complex 3D data has led to the development of high-dimensional frameworks for data learning. Here, we design several sparse learning-based approaches for high-dimensional data that effectively tackle multiple perception problems, including data filtering, data recovery, and data retrieval. The frameworks offer generative solutions for analyzing complex and irregular data structures without prior knowledge of data.
The first part of the dissertation proposes a novel method that simultaneously filters point cloud noise and outliers as well as completing missing data by utilizing a unified framework consisting of a novel tensor data representation, an adaptive feature encoder, and a generative Bayesian network. In the next section, a novel multi-level generative chaotic Recurrent Neural Network (RNN) has been proposed using a sparse tensor structure for image restoration. In the last part of the dissertation, we discuss the detection followed by localization, where we discuss extracting features from sparse tensors for data retrieval.Doctor of Philosophy
The development of automation systems and robotics brought the modern world unrivaled affluence and convenience. However, the current automated tasks are mainly simple repetitive motions. Tasks that require more artificial capability with advanced visual cognition are still an unsolved problem for automation. Many of the high-level cognition-based tasks require the accurate visual perception of the environment and dynamic objects from the data received from the optical sensor. The capability to represent, identify and interpret complex visual data for understanding the geometric structure of the world is 3D perception. To better tackle the existing 3D perception challenges, this dissertation proposed a set of generative learning-based frameworks on sparse tensor data for various high-dimensional robotics perception applications: underwater point cloud filtering, image restoration, deformation detection, and localization. Underwater point cloud data is relevant for many applications such as environmental monitoring or geological exploration. The data collected with sonar sensors are however subjected to different types of noise, including holes, noise measurements, and outliers. In the first chapter, we propose a generative model for point cloud data recovery using Variational Bayesian (VB) based sparse tensor factorization methods to tackle these three defects simultaneously. In the second part of the dissertation, we propose an image restoration technique to tackle missing data, which is essential for many perception applications. An efficient generative chaotic RNN framework has been introduced for recovering the sparse tensor from a single corrupted image for various types of missing data. In the last chapter, a multi-level CNN for high-dimension tensor feature extraction for underwater vehicle localization has been proposed.
39
Campagna, Anne. "Structural analysis of protein interaction networks." Doctoral thesis, Universitat Pompeu Fabra, 2012. http://hdl.handle.net/10803/84111.
Full textAbstract:
Interactions between proteins give rise to many functions in cells. In the lastdecade, highthroughput experiments have identified thousands of protein interactions, which are often represented together as large protein interaction networks. However, the classical way of representing interaction networks, as nodes and edges, is too limited to take dynamic properties such as compatible and mutually exclusive interactions into account. In this work, we study protein interaction networks using structural information. More specifically, the analysis of protein interfaces in threedimensional protein structures enables us to identify which interfaces are compatible and which are not. Based on this principle, we have implemented a method, which aims at the analysis of protein interaction networks from a structural point of view by (1) predicting possible binary interactions for proteins that have been found in complex experimentally and (2) identifying possible mutually exclusive and compatible complexes. We validated our method by using positive and negative reference sets from literature and set up an assay to benchmark the identification of compatible and mutually exclusive structural interactions. In addition, we reconstructed the protein interaction network associated with the G proteincoupled receptor Rhodopsin and defined related functional submodules by combining interaction data with structural analysis of the network. Besides its established role in vision, our results suggest that Rhodopsin triggers two additional signaling pathways towards (1) cytoskeleton dynamics and (2) vesicular trafficking.Las funciones de las proteínas resultan de la manera con la que interaccionan entre ellas. Los experimentos de alto rendimiento han permitido identificar miles de interacciones de proteínas que forman parte de redes grandes y complejas. En esta tesis, utilizamos la información de estructuras de proteínas para estudiar las redes de interacciones de proteínas. Con esta información, se puede entender como las proteínas interaccionan al nivel molecular y con este conocimiento se puede identificar las interacciones que pueden ocurrir al mismo tiempo de las que están incompatibles. En base a este principio, hemos desarrollado un método que permite estudiar las redes de interacciones de proteínas con un punto de vista mas dinámico de lo que ofrecen clásicamente. Además, al combinar este método con minería de la literatura y Los datos de la proteomica hemos construido la red de interacciones de proteínas asociada con la Rodopsina, un receptor acoplado a proteínas G y hemos identificado sus sub--‐módulos funcionales. Estos análisis surgieron una novel vıa de señalización hacia la regulación del citoesqueleto y el trafico vesicular por Rodopsina, además de su papel establecido en la visión.
40
Isik, Mehmet Talha. "Localization In Underwater Acoustic Sensor Networks." Master's thesis, METU, 2007. http://etd.lib.metu.edu.tr/upload/12608794/index.pdf.
Full textAbstract:
Underwater Acoustic Sensor Networks (UW-ASNs) have the potential to enable many applications such as environmental monitoring, undersea exploration and distributed tactical surveillance. In order to realize the potential gains of these applications, it is essential that the sensor nodes can be accurately located in a three dimensional underwater sensor network topology. Although many localization protocols
have been proposed recently for terrestrial sensor networks, the unique characteristics of the underwater acoustic communication channel, such as high and variable propagation delay, necessitate new localization protocols. In order to address this need, a localization protocol for UW-ASN, Three-Dimensional Underwater Localization (3DUL), is presented in this thesis. 3DUL achieves network-wide robust 3D localization by using a distributed and iterative algorithm. Importantly, 3DUL exploits only three surface buoys for localization. The sensor nodes leverage the low speed of sound to accurately determine the inter-node distances. We show through simulation experiments that the localization accuracy does not degrade significantly with an increase in the number of nodes, making 3DUL scalable.
41
Castelli, Filippo Maria. "3D CNN methods in biomedical image segmentation." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/18796/.
Full textAbstract:
A definite trend in Biomedical Imaging is the one towards the integration of increasingly complex interpretative layers to the pure data acquisition process. One of the most interesting and looked-forward goals in the field is the automatic segmentation of objects of interest in extensive acquisition data, target that would allow Biomedical Imaging to look beyond its use as a purely assistive tool to become a cornerstone in ambitious large-scale challenges like the extensive quantitative study of the Human Brain.
In 2019 Convolutional Neural Networks represent the state of the art in Biomedical Image segmentation and scientific interests from a variety of fields, spacing from automotive to natural resource exploration, converge to their development. While most of the applications of CNNs are focused on single-image segmentation, biomedical image data -being it MRI, CT-scans, Microscopy, etc- often benefits from three-dimensional volumetric expression.
This work explores a reformulation of the CNN segmentation problem that is native to the 3D nature of the data, with particular interest to the applications to Fluorescence Microscopy volumetric data produced at the European Laboratories for Nonlinear Spectroscopy in the context of two different large international human brain study projects: the Human Brain Project and the White House BRAIN Initiative.
42
Alja'afreh, Mohammad Mahmoud. "An Efficient Hybrid Objects Selection Protocol for 3D Streaming over Mobile Devices." Thèse, Université d'Ottawa / University of Ottawa, 2012. http://hdl.handle.net/10393/23602.
Full textAbstract:
With the rapid development in the areas of mobile manufacturing and multimedia communications, there is an increasing demand for Networked Virtual Environment (NVE) applications, such as Augmented Reality (AR), virtual walk-throughs, and massively multiplayer online games (MMOGs), on hand-held devices. Unfortunately, downloading and rendering a complex 3D scene is very computationally intensive and is not compatible with current mobile hardware specifications nor with available wireless bandwidth. Existing NVE applications deploy client/server based 3D streaming over thin mobile devices, which suffer from single point of failure, latency, and scalability issues. To address these issues, image-based rendering (IBR) and cloud-based 3D streaming have been introduced. The former introduces visual artifacts that reduce, and usually cancel, the realistic behaviors of the Virtual Environment (VE) application, while the latter is considered very expensive to implement. Peer-to-peer (P2P) 3D streaming is promising and affordable, but it has to tackle issues in object discovery and selection as well as content provider strategies. Distributing VE content over a mobile ad-hoc network (MANET) makes the system difficult to update due to the dynamic nature of the mobile clients. In order to tackle these issues, we came up with a novel protocol that combines the pros of both central and distributed approaches. Our proposed hybrid protocol, called OCTET, enables 3D scene streaming over thin devices in a way that can cope with current mobile hardware capabilities and mitigate the challenges of client/server and P2P 3D streaming. In fact, OCTET provides strategies that select, prioritize, and deliver only those objects that contribute to the user’s visible scene. OCTET is implemented using the "ns-2" simulation environment, and extensive experiments have clearly demonstrated significant achievements in mobile resource utilization, throughput, and system scalability.
43
Fay, Rebecca. "Feature selection and information fusion in hierarchical neural networks for iterative 3D-object recognition." [S.l. : s.n.], 2007. http://nbn-resolving.de/urn:nbn:de:bsz:289-vts-60447.
Full text
44
Opoku, Agyeman Michael. "Optimizing heterogeneous 3D networks-on-chip architectures for low power and high performance applications." Thesis, Glasgow Caledonian University, 2014. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.688307.
Full text
45
Zhao, Fangli. "Optical Control of "All Visible" Fluoroazobenzene-Containing Architectures: From Small Molecules to 3D Networks." Doctoral thesis, Humboldt-Universität zu Berlin, 2018. http://dx.doi.org/10.18452/19193.
Full textAbstract:
Ortho-Fluorazobenzole stellen eine der interessantesten Familien von Azobenzolen dar, die mit sichtbarem Licht geschaltet werden können. Seit ihrer ersten Erwähnung durch unsere Gruppe im Jahr 2012 wurden sie aufgrund ihrer hervorragenden photo/elektrochemischen Eigenschaften intensiv auf molekularer Ebene, für biologische Anwendungen und in Volumenmaterialien untersucht. Typischerweise können ortho-fluorierte Azobenzole in beide Richtungen mit sichtbarem Licht und hohem Photoumsatz geschaltet werden. Außerdem weisen die Z-Isomere überlegene thermische Halbwertszeiten (bis zu 2 Jahre) auf. In dieser Arbeit werden zwei Projekte vorgestellt, die auf unseren kürzlich erworbenen Kenntnissen über fluorierte Azobenzole basieren. Zunächst wurde ein gemischtes Azobenzoldimer dargestellt, welches komplementäre Absorptionsprofile sowie die leichte elektrochemische Isomerisierung ausnutzt und dadurch dessen vier Schaltzustände orthogonal adressiert werden können. Dieses wurde bezüglich seiner Photoisomerisierung, thermischen Relaxation und seines elektrochemischen Schaltverhaltens untersucht. Anschließend haben wir ein 3D-Polymernetzwerk durch kovalente Vernetzung einer polyethylenglykol(PEG)-basierten Vorstufe mit einem fluorierten Azobenzol hergestellt, was zur Bildung eines photoempfindlichen Hydrogels führte. Als Folge davon konnten die mechanischen Eigenschaften des Gels durch Bestrahlung mit sichtbarem Licht und der dadurch ausgelösten Azobenzol-Isomerisierung reversibel beeinflusst werden.Ortho-fluoroazobenzenes represent one of the most interesting family of visible-light-responsive azobenzenes. Since the first report by our group in 2012, they have been intensively studied at the molecular level, for biological applications, and in bulk materials, due to their outstanding photo/electrochemical properties. Typically, ortho-fluorinated azobenzenes can isomerize in both directions using visible light with high photo-conversions, and the Z-isomers exhibit superior thermal half-lives (up to 2 years). In this work, two projects based on our recently acquired knowledge of fluorinated azobenzenes are presented. First, exploiting complementary absorption profiles and ease of electrochemical isomerization, a mixed azobenzene dimer, whose four isomers can be orthogonally addressed was prepared. It was investigated from its photo-isomerization, thermal relaxation, and electrochemical isomerization aspects. Second, we prepared a photo-responsive hydrogel via covalently cross-linking a poly(ethylene glycol) (PEG)-based precursor with a fluorinated azobenzene forming a 3D polymer network. As a result, the gel’s mechanical properties could be tuned reversibly due to the azobenzenes’ isomerization triggered by visible light irradiation.
46
Almeida, Rodrigo de Matos Pires Tavares de. "3D terrain generation using neural networks." Master's thesis, 2020. http://hdl.handle.net/10071/22222.
Full textAbstract:
With the increase in computation power, coupled with the advancements in the field in the form of
GANs and cGANs, Neural Networks have become an attractive proposition for content generation. This
opened opportunities for Procedural Content Generation algorithms (PCG) to tap Neural Networks
generative power to create tools that allow developers to remove part of creative and developmental
burden imposed throughout the gaming industry, be it from investors looking for a return on their
investment and from consumers that want more and better content, fast. This dissertation sets out to
develop a PCG mixed-initiative tool, leveraging cGANs, to create authored 3D terrains, allowing users
to directly influence the resulting generated content without the need for formal training on terrain
generation or complex interactions with the tool to influence the generative output, as opposed to
state of the art generative algorithms that only allow for random content generation or are needlessly
complex. Testing done to 113 people online, as well as in-person testing done to 30 people, revealed
that it is indeed possible to develop a tool that allows users from any level of terrain creation
knowledge, and minimal tool training, to easily create a 3D terrain that is more realistic looking than
those generated by state-of-the-art solutions such as Perlin Noise.Com o aumento do poder de computação, juntamente com os avanços neste campo na forma de GANs e cGANs, as Redes Neurais tornaram-se numa proposta atrativa para a geração de conteúdos. Graças a estes avanços, abriram-se oportunidades para os algoritmos de Geração de Conteúdos Procedimentais(PCG) explorarem o poder generativo das Redes Neurais para a criação de ferramentas que permitam aos programadores remover parte da carga criativa e de desenvolvimento imposta em toda a indústria dos jogos, seja por parte dos investidores que procuram um retorno do seu investimento ou por parte dos consumidores que querem mais e melhor conteúdo, o mais rápido possível. Esta dissertação pretende desenvolver uma ferramenta de iniciativa mista PCG, alavancando cGANs, para criar terrenos 3D cocriados, permitindo aos utilizadores influenciarem diretamente o conteúdo gerado sem necessidade de terem formação formal sobre a criação de terrenos 3D ou interações complexas com a ferramenta para influenciar a produção generativa, opondo-se assim a algoritmos generativos comummente utilizados, que apenas permitem a geração de conteúdo aleatório ou que são desnecessariamente complexos. Um conjunto de testes feitos a 113 pessoas online e a 30 pessoas presencialmente, revelaram que é de facto possível desenvolver uma ferramenta que permita aos utilizadores, de qualquer nível de conhecimento sobre criação de terrenos, e com uma formação mínima na ferramenta, criar um terreno 3D mais realista do que os terrenos gerados a partir da solução de estado da arte, como o Perlin Noise, e de uma forma fácil.
47
Chiou, Wen-Hau, and 邱文顥. "Information Brokerage in 3D Wireless Sensor Networks." Thesis, 2013. http://ndltd.ncl.edu.tw/handle/58982491460790259409.
Full textAbstract:
碩士國立清華大學
資訊工程學系
101
In 3D wireless sensor networks, many connection-based information brokerage schemes, which use the node connection information rather than the geographic location in- formation, have been proposed. Each one, however, either cannot guarantee data retrieval or incurs a considerable amount of replication overhead costs, making these schemes hard to use in actual practice. In this thesis, a connection-based infor- mation brokerage schemes, RGIB3D, are proposed for 3D wireless sensor networks. RGIB3D are shown, using theoretical analysis, to be retrieval-guaranteed, and shown, with simulations, to have short retrieval latencies while ensuring moderate replication overhead costs in 3D wireless sensor networks.
48
Huang, Yang-Cheng, and 黃揚晟. "3D Point Cloud Registration Using Neural Networks." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/6ntjsu.
Full textAbstract:
碩士國立臺北科技大學
電機工程系
106
In order to enhance the performance of industrial production, manufacturing automation has been playing a critical role. Applications focused on automated 3D visual object detection, positioning, and identification are thus becoming important in manufacturing industry. Hence, 3D point cloud registration appears to be one of the core technologies. Therefore, some practical optimization methods are proposed in this thesis. Specifically, neural networks (NNs) and convolutional neural networks (CNNs) are actively integrated to accomplish 3D pose estimation. The training sample set for offline supervised learning is constructed by various geometric features and descriptors with labels based on randomly rotating and down-sampling the model point cloud. A number of optimized 3D point cloud registration architecture using both neural networks and ICP are proposed that make rigid alignment between the model point cloud and data point cloud which further determine the transformation between the two point clouds. The entire 3D point cloud registration task including sample set construction, training, registration, and inference has been successfully accomplished for typical objects. The proposed approaches have been validated by experiments.
49
Pascoal, Cláudio. "3D Convolutional Neural Networks for Identifying Protein Interfaces." Master's thesis, 2021. http://hdl.handle.net/10362/123467.
Full textAbstract:
Protein interaction is a fundamental part of nearly all biochemical processes and
proteins evolved specific surface regions for molecular recognition and interaction. These
regions are different from the remaining surface, with different amino acid compositions,
geometry and chemical properties. Detecting protein interfaces can lead to a better
understanding of protein interactions granting advantages to fields such as drug design
and metabolic engineering.
Most of the existing interface predictors use structured data, clearly defined data
types usually obtained from data sets. However, proteins are very complex molecules
and there is not a single property capable of distinguishing the interface from the rest of
the protein surface to all types of proteins. Indeed, deep learning arises as an adequate
approach able to capture feature from unstructured data as images, texts, sensor data
and volumes. In here, the aim was to identify interface regions in known protein spatial
structures together with their biochemical properties by exploring new applications of
3D convolutional neural networks.
For this, some state-of-the-art convolutional neural networks architectures were explored in order to find an architecture that suits this problem, and even more, have good
performance. Other state-of-the-art machine learning predictors are also considered to
identify the best biochemical properties to be added as new channels.
Afterward, the interface predictions will be compared with the ground-truth, obtained by calculating the distances of atoms between the different chains of the protein
complexes.A interação entre proteínas é fundamental em todos os processos biológicos e bioquímicos. As proteínas são compostas por regiões específicas que permitem o reconhecimento molecular e, consequentemente, interações com outras moléculas. Normalmente, estas regiões são estruturalmente diferentes da restante molécula sendo caracterizadas e compostas por aminoácidos diferentes, propriedades químicas e geometria diversa. A detecção das interfaces das proteínas pode ser uma mais valia no contexto de perceber a interação entre as mesmas e consecutivamente, ser vantajoso para o design de novos fármacos (ou drug design) e engenharia metabólica. As previsões de interfaces usam maioritariamente dados estruturados, ou seja, dados bem definidos normalmente obtidos em bancos de dados. No entanto, as proteínas são moléculas complexas o que impossibilita a distinção da sua interface, uma vez que não existe uma propriedade única e específica para todas. Deste modo, o deep learning é uma ferramenta fundamental porque usa características de dados não estruturados, como por exemplo a informação espacial da proteína, imagens, textos, dados de sensores ou volumes. O objetivo principal deste projeto é identificar regiões de interfaces através de estruturas tri-dimensionais de proteínas conhecidas juntamente com as respetivas distribuição espacial das suas propriedades, usando redes neuronais de convolução. Neste trabalho foram estudados algoritmos de deep learning para encontrar a rede neuronal mais adequada ao problema que pretendemos resolver com o melhor desempenho. Outros algoritmos de previsão foram considerados para identificar quais as melhores propriedades bioquímicas a serem usadas como novos canais de input. Seguidamente, as previsões do modelo foram comparadas com as interfaces reais, que foram obtidas pelo cálculo das distâncias dos átomos entre cadeias diferentes do mesmo complexo.
50
Chen, Yu-Chih, and 陳昱志. "A 3D-Mesh Scatternet Formation Method for Bluetooth Networks." Thesis, 2013. http://ndltd.ncl.edu.tw/handle/z4t8n2.
Full textAbstract:
碩士中華大學
電機工程學系碩士班
101
In this thesis, a 3D-Mesh formation method is proposed. The types of 3D-Mesh architecture were divided into triangular-shaped, quadrilaterals, pentagons and hexagons. The number of layers can be generated from level 1 to level 8. The type of its interconnection bridge contains RmS(Ring Inter-mesh Slave)、RMS(Ring Intra-mesh Slave)、and RmMS(Ring Mesh Slave). Scatternet formation procedures include two stages. The first stage is to select a BlueRoot, First, any two nodes with random probability to enter the inquiry or inquiry state and the inquiry node keeps the information of the other node. Then the inquiry node starts the probability decision again to enter either the inquiry or inquiry scan state. The step is repeated iteratively until the last node called BlueRoot is generated. In the second stage, BlueRoot computes the scatternet topology according to the predefined formula and passes the piconet information to the other masters. Each master pages and connects its slave and bridge to form its piconet until the scatternet is formed. The predefined formula can be deduced according to the number of nodes and the hierarchical relationship. In addition, an optimum method by the simulation results is designed to achieve the lowest hop length of 3D-Mesh topology. Computer simulation is performed to generate the performance metrics including the average hop length, throughput, packet delay, and packet delivery ratio. The performance results show that 3D-Mesh Network achieve good scatternet performance for the Bluetooth-based ad hoc networks.