Dissertations / Theses on the topic 'Underwater object detection'

Thesis (S.M. in Oceanographic Engineering)--Joint Program in Applied Ocean Science and Engineering (Massachusetts Institute of Technology, Dept. of Ocean Engineering; and the Woods Hole Oceanographic Institution), 1999.
Includes bibliographical references (leaves 85-89).
by Qiang Wang.
S.M.in Oceanographic Engineering

Thesis (MScEng)--Stellenbosch University, 2014.
ENGLISH ABSTRACT: This thesis describes the second project relating to the development of a SONAR (SOund Navigation And Ranging) object detection and collision avoidance system for use on an AUV (Autonomous Underwater Vehicle) at Stellenbosch University. The main goal is to develop and test techniques that make use of the existing SONAR laboratory platform and wide-beam SONAR transducers to detect and locate objects and their limits/bounds under water in the horizontal plane. The results of the work done show that it is possible to use wide-beam transducers to locate the centroid and edges of a flat target with an error that is significantly smaller than the beam-width. The techniques developed will enable the development of a cost-effective SONAR system that can be implemented on an AUV.
AFRIKAANSE OPSOMMING: Hierdie tesis beskryf the tweede projek rakende die ontwikkeling van ’n SONAR voorwerp opsporings en botsingvermydingstelsel vir gebruik op ’n OOV (Outonome Onderwater Voertuig) aan die Universiteit van Stellenbosch. Die hoofdoel is om tegnieke te ontwikkel en te toets wat gebruik maak van die bestaande SONAR laboratorium opstelling en wye-straal SONAR opnemers om die posisie van voorwerpe onder water te bepaal, sowel as die posisie van die voorwerp se rande in die horisontale vlak. Die resultate van die werk wat gedoen is wys dat dit moontlik is om wye-straal opnemers te gebruik om die posisie van die sentroïde en rande van ’n plat voorwerp te vind met ’n fout wat aansienlik kleiner is as die straal-wydte. Die tegnieke wat ontwikkel is sal ons in staat stel om ’n koste-effektiewe SONAR stelsel te ontwikkel wat op ’n OOV geïmplenteer kan word.

In recent decades, the rapid development of intelligent vehicle and 3D scanning tecnologies has led to a growing interest in applications based on 3D point data processing, with many applications such as augmented reality or robot manipulation and obstacle avoidance, scene understanding, robot navigation, tracking and assistive technology among others, requiring an accurate solution for the 3D pose of the recognized objects. Thus object recognition is becoming an important topic in computer vision, where machine vision and robotics techniques are becoming key players. In this thesis work, the main objective is to develop a semantic mapping method by integrating a 3D object recognition pipeline with a feature-based SLAM system, in order to assist autonomous underwater interventions in the near future. To this end, the work proposed in this paper targets three axes. First, it aims to compare the performance of 3D global descriptors within the state of the art, focusing on those based on point clouds and targeted at real-time object recognition applications. For this purpose, we selected a set of test objects representative of Inspection, Maintenance and Repair (IMR) applications and whose shape is usually known a priori. Their CAD models were used to: 1) create a data base of synthetic object views used as a priori knowledge, and 2) simulate the point clouds that would be gathered during the scanning under realistic conditions, with added noise and varying resolution. Extensive experiments were performed with both virtual scans and real data collected with an AUV equipped with a fast laser scanner developed at our research centre. The second goal of our work was to use a real-time laser scanner mounted on an AUV to detect, identify, and locate objects in the robot’s environment, with the aim of allowing an intervention Autonomous Underwater Vehicle (I-AUV) to know what manipulation actions could be performed on each object. This goal was tackled by the design and development of a 3D object recognition method for uncolored point clouds (laser scans) using point features. The algorithm uses a database of partial views of the objects stored as point clouds. The recognition pipeline includes 5 stages: 1) Plane segmentation, 2) Pipe detection, 3) Semantic Object-segmentation, 4) Feature-based Object Recognition and 5) Bayesian estimation. To apply Bayesian estimation, it is necessary to track objects across scans. For this purpose, the Inter-distance Joint Compatibility Branch and Bound (IJCBB) data association algorithm was proposed based on the distances between objects. The performance of the method was tested using a dataset of the inspection of a pipe infrastructure made of PVC objects connected by pipes. The structure is representative of those commonly used by the offshore industry. Experimental results show that Bayesian estimation improves the recognition performance with respect to the case where only the descriptor is used. The inclusion of semantic information about object pipe connectivity further improves recognition performance. The final goal of the thesis, consists of integrating the 3D object recognition system with a feature-based SLAM system to implement a semantic map providing the robot with information about the location and the type of objects in its surroundings. The SLAM improved both the accuracy and reliability of pose estimates of the robot and the objects. This is especially important in challenging scenarios where significant changes in viewpoint and appearance arise
A les darreres dècades, el ràpid desenvolupament de vehicles intel·ligents i de les tecnologies d’escaneig 3D han contribuït a augmentar l’interès en les aplicacions basades en processament de núvols de punts 3D, amb aplicacions com la realitat augmentada, la manipulació robòtica, l’evasió d’obstacles, la comprensió d’escenes, la navegació robòtica, el seguiment d’objectes i la tecnologia d’assistència, etc., que requereixen una soluci´o precisa de la posició 3D i l’orientació d’un objecte. Per tant, el reconeixement d’objectes s’està convertint en un tema, on la visió per computador i les tècniques robòtiques esdevenen protagonistes clau. En aquest treball de tesi, l’objectiu principal és desenvolupar un mètode per a la construcció de mapes semàntic mitjançant la integració d’una cadena de processament per al reconeixement d’objectes 3D, amb un sistema de SLAM basat en característiques, amb l’objectiu d’ajudar a les futures intervencions submarines. Per això, el treball proposat en aquesta tesi es divideix en tres eixos principals. El primer té com a objectiu comparar el rendiment de descriptors globals d’última generació, centrant-se en els basats en núvols de punts 3D i destinats a aplicacions de reconeixement d’objectes en temps real. Per a aquest objectiu, s’ha seleccionat un conjunt d’objectes de prova representatius d’aplicacions d’inspecció, manteniment i reparació (IMR), la forma dels quals es coneix a priori. Els seus models CAD s’han utilitzat per a: 1) crear una base de dades amb les vistes sintètiques dels objectes, i 2) simular els núvols de punts que adquiriria, en condicions realistes, un escàner làser incloent soroll sintètic i simulant diferents resolucions. S’han dut a terme experiments tant a partir d’escaneigs virtuals com de dades reals recopilades amb un AUV equipat amb un escàner làser de temps real desenvolupat al nostre centre de recerca. El segon objectiu del nostre treball va consistir en utilitzar aquest escàner làser, muntat a un AUV per detectar, reconèixer i localitzar objectes a l’entorn del robot, per tal de permetre, a un Vehicle Submarí Autònoms d’Intervenció (IAUV), saber quines accions de manipulació podria fer amb cada objecte. Aquest objectiu es va abordar amb el disseny i el desenvolupament d’un mètode de reconeixement d’objectes 3D en núvols de punts incolors (escanejos làser) utilitzant descriptors dels punts 3D. L’algorisme utilitza una base de dades de vistes parcials dels objectes emmagatzemats en forma de núvols de punts. El procés de reconeixement consta de 5 passos: 1) Segmentació de plànols, 2) Detecció de canonades, 3) Segmentació semàntica d’objectes, 4) Reconeixement d’objectes a partir dels descriptors de punts 3D i 5) Estimació bayesiana. Per aplicar l’estimació bayesiana, cal ser capaços de fer un seguiment dels objectes en escanejos successius. Per fer-ho, s’ha proposat l’algorisme Inter-distance Joint-Compatibility Branch and Bound (IJCBB) d’associació de dades basada en les distancies entre objectes dins del núvol de punts. El rendiment del mètode es va avaluar fent servir dades experimentals relatives a la inspecció d’una infraestructura composta de canonades interconnectades per objectes de PVC. L’estructura ´es representativa de les comunament utilitzades per la indústria offshore. Els resultats experimentals mostren que l’estimació bayesiana millora el rendiment del reconeixement en comparació de l’ús ´únic del descriptor. La inclusió d’informació semàntica sobre la connectivitat d’objectes a canonades millora encara més el rendiment del reconeixement. L’objectiu final de la tesi va abordar la integració del sistema de reconeixement d’objectes 3D basat en descriptors amb un sistema de SLAM basat en característiques, per implementar un mapa semàntic que proporciona al robot informació sobre la ubicació i el tipus d’objectes a l’entorn. La utilització de tècniques de SLAM ha millorat la precisió i la fiabilitat de les estimacions de la postura del robot i els objectes. Això és especialment important en escenaris difícils on es produeixen canvis significatius de perspectiva i aparença
Programa de Doctorat en Tecnologia

Seagrasses play an essential role in the marine ecosystem by providing foods, nutrients, and habitat to the marine lives. They work as marine bioindicators by reflecting the health condition of aquatic environments. Seagrasses also act as a significant atmospheric carbon sink that mitigates global warming and rapid climate changes. Considering the importance, it is critical to monitor seagrasses across the coastlines which includes detection, mapping, percentage cover calculation, and health estimation. Remote sensing-based aerial and spectral images, acoustic images, underwater two-dimensional and three-dimensional digital images have so far been used to monitor seagrasses. For close monitoring, different machine learning classifiers such as the support vector machine (SVM), the maximum likelihood classifier (MLC), the logistic model tree (LMT) and the multilayer perceptron (MP) have been used for seagrass classification from two-dimensional digital images. All of these approaches used handcrafted feature extraction methods, which are semi-automatic. In recent years, deep learning-based automatic object detection and image classification have achieved tremendous success, especially in the computer vision area. However, to the best of our knowledge, no attempts have been made for using deep learning for seagrass detection from underwater digital images. Possible reasons include unavailability of enough image data to train a deep neural network. In this work, we have proposed a Faster R-CNN architecture based deep learning detector that automatically detects Halophila ovalis (a common seagrass species) from underwater digital images. To train the object detector, we have collected a total of 2,699 underwater images both from real-life shorelines, and from an experimental facility. The selected seagrass (Halophila ovalis) are labelled using LabelImg software, commonly used by the research community. An expert in seagrass reviewed the extracted labels. We have used VGG16, Resnet50, Inception V2, and NASNet in the Faster R-CNN object detection framework which were originally trained on COCO dataset. We have applied the transfer learning technique to re-train them using our collected dataset to be able to detect the seagrasses. Inception V2 based Faster R-CNN achieved the highest mean average precision (mAP) of 0.261. The detection models proposed in this dissertation can be transfer learned with labelled two-dimensional digital images of other seagrass species and can be used to detect them from underwater seabed images automatically.

Thesis (S.M.)--Joint Program in Oceanography/Applied Ocean Science and Engineering (Massachusetts Institute of Technology, Dept. of Mechanical Engineering; and the Woods Hole Oceanographic Institution), 2009.
Includes bibliographical references (p. 88-91).
The work presented in this thesis addresses the problem of the detection, classification and localization of seabed objects in shallow water environments using a time reversal approach in a bistatic configuration. The waveguide is insonified at low frequency ('kHz) with an omnidirectional source and the resulting scattered field is sampled by a receiving array towed behind an Autonomous Underwater Vehicle (AUV). The recorded signals are then processed to simulate onboard the AUV, the time reversed transmissions which serve to localize the origin of the scattered field on the seabed and estimate the position of the targets present. The clutter rejection based upon the analysis of the singular values of the Time Reversal operator is investigated with simulated data and field measurements collected off the coast of Palmaria (Italy) in January 2008.
by Alexis J. Dumortier.
S.M.

Les mines sous marines sont très utilisées dans les conflits. Pour contrer cette menace, les marines s'équipent de moyens de lutte anti-mine autonomes afin d'éviter l'intervention humaine. Une mission de guerre des mines se découpe en quatre étapes distinctes : la détection des objets, la classification et l'identification puis la neutralisation. Cette thèse propose des solutions algorithmiques pour l'étape d'identification par caméra vidéo. Le drone d'identification connaît la position approximative de l'objet à identifier. La première mission de ce drone est de re-détecter l'objet avant de le classifier et de l'identifier. Le milieu sous-marin perturbe les images acquises par la caméra (absorption, diffusion). Pour faciliter la détection et la reconnaissance (classification et identification), nous avons prétraité les images. Nous avons proposé deux méthodes de détection des objets. Tout d'abord nous modifions le spectre de l'image afin d'obtenir une image dans laquelle il est possible de détecter les contours des objets. Une seconde méthode a été développée à partir de la soustraction du fond, appris en début de séquence vidéo. Les résultats obtenus avec cette seconde méthode ont été comparés à une méthode existante. Lorsqu'il y a une détection, nous cherchons à reconnaître l'objet. Pour cela, nous utilisons la corrélation. Les images de référence ont été obtenues à partir d'images de synthèse 3D des mines. Pour les différentes méthodes utilisées, nous avons optimisés les résultats en utilisant les informations de navigation. En effet, selon les déplacements du drone, nous pouvons fixer des contraintes qui vont améliorer la détection et réduire le temps de calcul nécessaire à l'identification.

碩士
國立中山大學
海洋環境及工程學系研究所
104
Disaster or accident happened in or related to the aquatic environment is much more difficulty and dangerous to the rescuers as compared to those happened in the terrestrial environment. In this circumstance, detection, location and salvage the victims under the water or on the water bottom are the major challenges faced by the rescuers. An efficient and reliable way of detecting and locating the victims is therefore a major component of the operation. After the TransAsia Airways flight GE235 plane crash landed in Keelung River in February, 2015 (codenamed Nan-Kong Aviation Accident), 12 people were missing and were expected to have been pulled down the river. After exhausted searching operation by divers but in vain, the potential location of the 3 still missing people were seriously re-examined. It was concluded that the Nan-Hu bridge area is one of the most potential areas and a comprehensive underwater acoustic investigation of this specific area was recommended. A bottom fixed mechanically scanned imaging sonar was therefore employed for this specific searching for operation. Based on underwater acoustic images collected, underwater environment and topographic features of the Nan-Hu bridge area can be divided into six distinctive regions, in a sequence from upstream to downstream, i.e., flow channel region, upstream region, submerged dike region, depositional region, erosion region, and downstream region. Based on outer appearance, composition and dimensions, a total of 25 potential or unidentified targets were recognized in the high resolution acoustic images. The 25 potential targets positioned in Nan-Hu bridge area were further identified and proved individually by divers. In this case, two of the 3 unaccounted personnel were located at the submerged dike region. All the other potential targets were proved to be either concrete blocks or logs. In addition, it was also concluded that the last unaccounted personnel was not located in this specific bridge area. From a time series point of view, the dates that the unaccounted personnel were detected and salvaged in this accident could be grouped into two categories. At flat and muddy bottom regions, searching for operations by divers could be completed effectively and comprehensively in a short period of time. However, prominent technical limitations existed for locating those missing personnel by divers at the rugged regions which might detain the progress of the whole searching for operation.

碩士
國立中山大學
海洋環境及工程學系研究所
101
Search for drowned human body by divers is a highly technical dependent, difficult and dangerous work. How to improve this kind of underwater rescue work, under the considerations of the safety of divers and the effectiveness in detection of the target within a limited period of time, is a major subject for fire fighters. For the time being, a normal search and rescue procedure basically includes visual observation of the bank area and surficial water area. In addition, divers are sometimes sent out to complement the search procedure by either visual observation or tactile. A much more efficient way to conduct this kind of recovery activity is to incorporate the state of art of the underwater acoustics technique, such as scanning sonar, into the operation procedures. This investigation was focused on the application of scanning sonar and image analysis techniques as well as seafloor object identification skills for the detection of drowned human body. In addition, safety of divers under water and their activities could be improved and monitored. Remotely operated vehicle (ROV) and towed operated vehicle (TOV) should be incorporated into the normal search procedure for the purpose of improving target identification in the future. Under this circumstance, both target searching rate and divers’ safety could be effectively improved or guaranteed. The proposed procedure which incorporated both acoustical (i.e., scanning sonar) and optical (i.e., ROV or TOV) apparatus are expected to simplify and improve the underwater target search and identification activities and will allow fire fighters a more professional and safety way in conducting drowned human body recovery activities.

The exploration of oceans and sea beds is being made increasingly possible through the development of Autonomous Underwater Vehicles (AUVs). This is an activity that concerns the marine community and it must confront the existence of notable challenges. These include, for example, mining minerals, inspecting pipeline and mapping oceans, sampling in contaminated water. Also, there has been another growing interest for security forces in precluding submarines or intruders from a beach or harbour entrance as well as hunting shallow water mines. However, an automatic detecting and tracking system is the first and foremost element for an AUV or an aqueous surveillance network. Since accurate surrounding information is essential in order to manoeuvre the AUV efficiently and economically, while corrupt information can jeopardize an entire mission. By extracting the space information form sensors, an AUV can achieve the localisation and mapping which are currently two primary concerns in the robotics research. Meanwhile, such information will provide a fundament of protection for surface vessels or troops, harbour infrastructure and oil plant against the enemy and terrorism. Acoustic sensors are commonly used to detect and position underwater obstacles, suspicious objects or to map the surroundings because sound waves can propagate more appreciable distances than electromagnetic and optical energy in the water. The measurements from these sensors, however, are always bound up with noises and errors. Various underwater activities may further pollute sound signals and then threaten the AUV navigation process. To simplify the detection procedure, some researchers make use of acoustic beacons or apparent obstructions (such as rocks, concrete walls) because they have distinctive characteristics. Point or line features are extracted from the acoustic signals or images for localization and mapping purposes. The long propagation range of sound waves can present new problems when acoustic sensors operate in confined environments, such as water tanks, rivers and harbours. The multiple reflections will be recorded by the sensor and result in false alarms. Furthermore, with advances in manufacturing techniques, the downsizing in marine explosive ordnances is progressing significantly, making it more difficult to discriminate between surface reflections and explosive ordnances. Finally, under the consideration of cost effectiveness, a mechanically scanned sonar has been introduced for the AUV in this research. However, the sensor beam cannot cover a large region simultaneously and a moving object may be distorted in the acoustic image because of the relatively low scanning speed. Due to such distortions in the data flows, objects may be indistinguishable from random noise or reverberation in acoustic images. The research presented here addresses the afore-mentioned problems relating to the theme of automatic detection from acoustic images. It is concerned with the detection and tracking of small underwater objects in order to protect autonomous underwater vehicles using sonar (SOund Navigation and Range). In the present study, these vehicles operated in laboratory water tanks or natural river environments. This research made use of self provided analytical studies that differentiated between reverberation and real object echoes. Detections were achieved automatically by using signal and image processing techniques. This research consists of three important and linked strategies. Firstly, a simple and fast reverberation suppression filter was provided, based on the understanding of the mechanism of the sonar sensor. Secondly, a robust detection system was developed to perceive small suspended obstacles in the water. Thirdly and finally, arc features were successfully extracted from the acoustic images and mathematical maps were generated from those features. The majority of experiments were derived from the elliptical water tank and the River Torrens, Adelaide, South Australia. For this project, a sequence of sonar images was taken from the same sonar location in the elliptical water tank. Further, a sequence of sonar images was taken from a sequence of sonar locations in the natural river. They provided different data sets for the assessment and evaluation of self developed algorithms. Results shown in this thesis confirm the favourable outcomes of the investigation and applied methodology.
http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1454839
Thesis (M.Eng.Sc.) -- University of Adelaide, School of Mechanical Engineering, 2010

The exploration of oceans and sea beds is being made increasingly possible through the development of Autonomous Underwater Vehicles (AUVs). This is an activity that concerns the marine community and it must confront the existence of notable challenges. These include, for example, mining minerals, inspecting pipeline and mapping oceans, sampling in contaminated water. Also, there has been another growing interest for security forces in precluding submarines or intruders from a beach or harbour entrance as well as hunting shallow water mines. However, an automatic detecting and tracking system is the first and foremost element for an AUV or an aqueous surveillance network. Since accurate surrounding information is essential in order to manoeuvre the AUV efficiently and economically, while corrupt information can jeopardize an entire mission. By extracting the space information form sensors, an AUV can achieve the localisation and mapping which are currently two primary concerns in the robotics research. Meanwhile, such information will provide a fundament of protection for surface vessels or troops, harbour infrastructure and oil plant against the enemy and terrorism. Acoustic sensors are commonly used to detect and position underwater obstacles, suspicious objects or to map the surroundings because sound waves can propagate more appreciable distances than electromagnetic and optical energy in the water. The measurements from these sensors, however, are always bound up with noises and errors. Various underwater activities may further pollute sound signals and then threaten the AUV navigation process. To simplify the detection procedure, some researchers make use of acoustic beacons or apparent obstructions (such as rocks, concrete walls) because they have distinctive characteristics. Point or line features are extracted from the acoustic signals or images for localization and mapping purposes. The long propagation range of sound waves can present new problems when acoustic sensors operate in confined environments, such as water tanks, rivers and harbours. The multiple reflections will be recorded by the sensor and result in false alarms. Furthermore, with advances in manufacturing techniques, the downsizing in marine explosive ordnances is progressing significantly, making it more difficult to discriminate between surface reflections and explosive ordnances. Finally, under the consideration of cost effectiveness, a mechanically scanned sonar has been introduced for the AUV in this research. However, the sensor beam cannot cover a large region simultaneously and a moving object may be distorted in the acoustic image because of the relatively low scanning speed. Due to such distortions in the data flows, objects may be indistinguishable from random noise or reverberation in acoustic images. The research presented here addresses the afore-mentioned problems relating to the theme of automatic detection from acoustic images. It is concerned with the detection and tracking of small underwater objects in order to protect autonomous underwater vehicles using sonar (SOund Navigation and Range). In the present study, these vehicles operated in laboratory water tanks or natural river environments. This research made use of self provided analytical studies that differentiated between reverberation and real object echoes. Detections were achieved automatically by using signal and image processing techniques. This research consists of three important and linked strategies. Firstly, a simple and fast reverberation suppression filter was provided, based on the understanding of the mechanism of the sonar sensor. Secondly, a robust detection system was developed to perceive small suspended obstacles in the water. Thirdly and finally, arc features were successfully extracted from the acoustic images and mathematical maps were generated from those features. The majority of experiments were derived from the elliptical water tank and the River Torrens, Adelaide, South Australia. For this project, a sequence of sonar images was taken from the same sonar location in the elliptical water tank. Further, a sequence of sonar images was taken from a sequence of sonar locations in the natural river. They provided different data sets for the assessment and evaluation of self developed algorithms. Results shown in this thesis confirm the favourable outcomes of the investigation and applied methodology.
Thesis (M.Eng.Sc.) -- University of Adelaide, School of Mechanical Engineering, 2010

碩士
國立成功大學
系統及船舶機電工程學系碩博士班
101
The purpose of this paper is developing a procedure to distinguish target characteristics such as material and size by observing convergence frequency through iterative time reversal process. The proposed procedure is verified from designed simulations and experiments by using slim shell circular tubes as targets. These slim shell structures produce lamb waves that contain energy within corresponding frequency bandwidth. The iterative time reversal process is then performed to enhance energy of responding lamb waves and makes it possible to distinguish target characteristics. In addition, target distance can be determined from wave propagation time. In experiments, interrogating signals are controlled by LabVIEW and emitted from an unfocused broadband underwater transducer. Specular echo and responding lamb wave signals are then received by another same type transducer and recorded by a digital oscilloscope. Due to the enhancing effect of iterative time reversal, energy of iterative receiving signals converges to certain frequencies. After comparing these frequencies from varying target material and target size, the result proves that the proposed procedure is capable to distinguish properties of different targets. The simulation of a thin and fluid-loaded elastic shell is computed using the classical theoretical formulation of Goodman and Stern to create shell model and find convergence frequency