Dissertations / Theses on the topic 'Crowd Flow'

Crowd simulations are being used in an increasing number of different applications, like evacuation scenarios, video games and movie special effects This creates a demand for crowd simulators that are simple to use and accessible to users of varying backgrounds. We will study the flow tile method proposed by Chenney [1], which provides an intuitive way of interactively designing divergence free velocity fields for various applications. A reimplementation of Chenney's method will be given and the implementation will be evaluated in terms of user-friendliness and how well the use of static spatially defined velocity fields suits crowd simulation. Furthermore the possibility of using the velocity fields for other related applications such as mobile robotics will be touched on as well.
Simuleringar av folkmassor används i ett ökande antal olika tillämpningar, som evakueringsscenarion, datorspel och speciale­ffekter för film. Detta skapar en efterfrågan efter simulatorer som är enkla att använda och tillgängliga för användare från olika ämnesområden och bakgrunder. Vi kommer att studera flow tile-metoden som Chenney [1] föreslår. Metoden är ett intuitivt och interaktivt sätt att skapa divergensfria hastighetsfält för olika tillämpningar. En omimplementation av Chenneys metod kommer att ges och implementationen kommer att evalueras i termer av användarvänlighet och hur väl användningen av hastighetsfält som är statiska och definierade i rummet passar för simulering av folkmassor. Vidare kommer möjligheten att använda hastighetsfälten för andra liknande tillämpningar, som robotik, att diskuteras också.

Crowd simulation is a technique used to model real people as agents in a computer-generated environment. Generating simulated crowds can help the research process of testing with agents in multiple scenarios without having to use real people. The flow of large crowds in different space-designs vary and this thesis will examine flow rates in funnel-shaped constructions and bottleneck constructions with the help of computer-generated crowd simulations. These flows can impact individuals to the extent where they are hurt by misdirected forces. Two different variables are defined for the funnel-shaped construction: corridor width and funnel-angle. For the bottleneck construction, there is only a corridor width property. Agents moving through an environment only move in one direction with no possibilities of a headon collision. Timing the agents that walk through the environments and looking at the forces they form shows that the funnels had a better flow and took less time than the bottlenecks. When looking at the different widths of the corridor the wider constructions was faster than all of the narrower constructions except the case with the narrow 15-degree funnel that was faster than the wide bottleneck. Introducing a funnel-shaped construction shows an improvement of the flow rate where you get better results the lower the angles. Applications of the funnel-shaped construction can include urban planning and architecture for spaces that have a capacity for large crowds, where an improvement in flow rate can make the space safer in an evacuation.
Simulation av folkmassor är en teknik som används för att modellera riktiga personer i en datorgenererad miljö som agenter. Att generera simulerade folkmassor kan hjälpa till med forskningsprocessen för testning av agenter i flertal scenarion utan att behöva använda riktiga personer. Flödet för folkmassor kan i hög grad variera för olika rumskonstruktioner. I den här avhandlingen undersöks flödeshastigheten i flaskhals- och trattformadekonstruktioner med hjälp av datorgenererade simulationer. Flödena kan påverka individer till den grad så att de skadas av missriktade krafter. Två variabler för de trattformade konstruktionerna är definerade, där den ena är bredden på öppningen till korridoren och den andra är vinkeln på tratten. För flaskhalskonstruktionerna används endast variabeln för bredden. Agenter i konstruktionen rör sig i samma riktning så det finns inte någon risk för frontalkollision. Resultat från tidtagning och kraftpilarnas riktning när agenterna går igenom konstruktionerna visar att en trattformad konstruktion ger ett bättre flöde samt tar mindre tid. En flaskhalskonstruktion med en bredare öppning ger bättre resultat än alla konstruktioner med en smalare öppning förutom den med 15-gradig vinkel på tratten. Att introducera en trattform till konstruktionen visar på en förbättrad flödeshastighet där lägre trattvinklar ger bättre resultat. Applikationer av den trattformade konstruktionen kan inkludera stadsplanering och arkitektur för rum som har en kapacitet för stora folkmassor, där en förbättring av flödeshastighet kan göra rummen sakrare i en evakuering.

Novel computer vision techniques have been developed for automatic monitoring of crowed environments such as airports, railway stations and shopping malls. Using video feeds from multiple cameras, the techniques enable crowd counting, crowd flow monitoring, queue monitoring and abnormal event detection. The outcome of the research is useful for surveillance applications and for obtaining operational metrics to improve business efficiency.

The Particle Image Velocimetry optical flow algorithm had better correspondence with the velocity vectors of manually tracked crowd members than results obtained using the Lukas-Kanade method. Here, also, it was found that 40x40 pixel cells were better than 15x15. A second attempt at quantifying the correlation between optical flow and actual crowd member velocities was studied using simulations. Two processes were researched, which utilized geometrical correction of the perspective distortion of the crowd videos. One process geometrically corrects the video, and then obtains optical flow data. The other obtains optical flow data from video, and then geometrically corrects the data. The results indicate that the first process worked better. Correlation was calculated between sets of data obtained from the average of twenty frames. This was found to be higher than calculating correlations between the velocities of cells in each pair of frames. An experiment was carried out to predict crowd tracks using optical flow and a calculated parameter, beta, seems to give promising results.; The need exists for finding a quantitative method for validating crowd simulations. One approach is to use optical flow of videos of real crowds to obtain velocities that can be used for comparison to simulations. Optical flow, in turn, needs to be calibrated to be useful. It is essential to show that optical flow velocities obtained from crowd videos can be mapped into the spatially averaged velocities of the observed trajectories of crowd members, and to quantify the extent of the correlation of the results. This research investigates methods to uncover the best conditions for a good correlation between optical flow and the average motion of individuals in crowd videos, with the aim that this will help in the quantitative validation of simulations. The first approach was to use a simple linear proportionality relation, with a single coefficient, alpha, between velocity vector of the optical flow and observed velocity of crowd members in a video or simulation. Since there are many variables that affect alpha, an attempt was made to find the best possible conditions for determining alpha, by varying experimental and optical flow settings. The measure of a good alpha was chosen to be that alpha does not vary excessively over a number of video frames. Best conditions of low coefficient of variation of alpha using the Lucas-Kanade optical flow algorithm were found to be when a larger aperture of 15x15 pixels was used, combined with a smaller threshold. Adequate results were found at cell size 40x40 pixels; the improvement in detecting details when smaller cells are used did not reduce the variability of alpha, and required much more computing power. Reduction in variability of alpha can be obtained by spreading the tracked location of a crowd member from a pixel into a rectangle.
ID: 030423090; System requirements: World Wide Web browser and PDF reader.; Mode of access: World Wide Web.; Thesis (M.S.)--University of Central Florida, 2011.; Includes bibliographical references (p. 104-108).
M.S.
Masters
Sciences
Modeling and Simulation

Within the design of any building, there is a requirement for designers to understand the intended purposes of the building and the elements that influence performance. These elements can be as tangible as providing a lecture hall within a university or relatively intangible such as the environmental temperatures of the rooms. The elements involved are generally recognised within the design industry and a combined force of engineers, architects, and specialist advisors work together to ensure all of the elements are in place for each new design. However, one element affecting performance that has not yet been comprehensively covered (at least for many building types) is that relating to occupant movement and the influence this has on experience and hence performance. For example, the number of times people have to negotiate cross-flow environments in a train station before becoming agitated is unknown. Also, the average distance people will walk through a shopping centre before becoming tired and ending the activity is unknown. Even so, they will both be impacted upon by the design and they will both reflect back on the performance of the design. Before starting this research, it was realised by the research engineer that there was only a limited understanding and application of people flow analyses within industry and, where it existed, it was solely related to transport terminals, pedestrian walkways/crossings, sports stadia arrivals/egress, and evacuation analyses.

Nous étudions dans ce travail les mouvements de foule intervenant dans les situa- tions d'urgence. Nous proposons un modèle macroscopique (la foule est représentée par une densité de personnes) obéissant à deux principes très simples. Tout d'abord, chaque personne possède une vitesse souhaitée (typiquement celle qui la mène vers la sortie), qu'elle adopterait en l'absence des autres. Ensuite, la foule doit respecter une contrainte de congestion, et la densité de personnes doit rester inférieure à une valeur fixée. Cette contrainte impose une vitesse de déplacement différente de la vitesse souhaitée. Nous choisissons de prendre comme vitesse réelle celle qui est la plus proche, au sens des moindres carrés, de la vitesse souhaitée, parmi les champs de vitesses admissibles, au sens où ils respectent la contrainte de densité maximale. Le modèle obtenu s'écrit sous la forme d'une équation de transport impliquant une vitesse peu régulière a priori, et qui ne peut être étudiée par des méthodes classiques. Nous démontrons un résultat d'existence grâce à la théorie du transport optimal, tout d'abord dans le cas d'une vitesse donnée comme le gradient d'une fonction, puis dans le cas général. Nous mettons également en œuvre un schéma numérique de type catching-up : à chaque pas de temps, la densité est déplacée selon le champ de vitesse souhaitée, puis est projetée sur l'ensemble des densités admissibles. Les résultats obtenus fournissent des temps d'évacuation dont l'ordre de grandeur est proche de la réalité.

Computer hardware and algorithm design have seen significant progress over the years. It is also seen that there are several domains in which humans are more efficient than computers. For example in image recognition, image tagging, natural language understanding and processing, humans often find complicated algorithms quite easy to grasp. This thesis presents the different case studies to learn human mapping strategy to solve the mapping problem in the area of coarse-grained reconfigurable architectures (CGRAs). To achieve optimum level performance and consume less energy in CGRAs, place and route problem has always been a major concern. Making use of human characteristics can be helpful in problems as such, through pattern recognition and experience. Therefore to conduct the case studies a computer mapping game called UNTANGLED was analyzed as a medium to convey insights of human mapping strategies in a variety of architectures. The purpose of this research was to learn from humans so that we can come up with better algorithms to outperform the existing algorithms. We observed how human strategies vary as we present them with different architectures, different architectures with constraints, different visualization as well as how the quality of solution changes with experience. In this work all the case studies obtained from exploiting human strategies provide useful feedback that can improve upon existing algorithms. These insights can be adapted to find the best architectural solution for a particular domain and for future research directions for mapping onto mesh-and- stripe based CGRAs.

Organic synthesis often requires one functional group of a bifunctionalised compound to be rendered temporarily inert to allow the selective reaction of another moiety. While protecting groups are used to remove the problem of the functional group incompatibility, they also raise other issues such as increasing the length of the synthetic pathway (by at least two steps – protection and deprotection), generally leading to an increase in cost and a decrease in yield. The protecting group is often selected based upon the deprotection conditions, leading to the requirement for orthogonal protecting groups. Crown ethers are commonly associated with the complexation of metal ions, but the 18-crown-6 species also readily bind ammonium ions with complexation occurring via hydrogen bonding. As discussed in Chapter 1, crown ethers have previously been employed for N-protection in this way, to successfully facilitate the reaction of bifunctional compounds, though they have exhibited very little selectivity and reaction control. Isolation of the desired product from the resulting reaction mixture has also proven to be problematic.Over recent years there has been a large increase in the volume of organic protocols conducted in micro and continuous flow reactors. Utilising the high surface to volume ratio obtained under these reaction conditions, greater reaction control of many common and specialised organic syntheses has been reported. Building upon literature precedent, the work herein reports the immobilisation of an 18-crown-6 ether derivativeonto a solid-support and its incorporation into a continuous flow reactor to enable sequestration of the primary amine salt of a bi-functionalised compound. This effectively affords a non-covalent N-protection strategy allowing the selective reaction of the remaining moiety. The desired product is subsequently recovered as the free amine by a simple process of decomplexation using an organic base.

Orientadores: Siome Klein Goldenstein, Jacques Wainer
Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação
Made available in DSpace on 2018-08-22T23:53:33Z (GMT). No. of bitstreams: 1 Coelho_RenatoSchattanPereira_M.pdf: 864445 bytes, checksum: 8f57902047a23925af4b81fa0d7f3188 (MD5) Previous issue date: 2013
Resumo: O trânsito é um problema cada vez maior nas cidades, consumindo recursos e agravando a poluição; em São Paulo perdem-se cerca de R$33 bilhões por ano por causa do trânsito. Neste trabalho de mestrado desenvolvemos um sistema que une as áreas de simulação de multidões e planejamento probabilístico para otimizar semáforos de tempo fixo. Essas duas áreas apresentam algoritmos que permitem soluções eficientes para os problemas, mas a sua aplicação ainda depende largamente da intervenção de especialistas no problema a ser estudado, seja descrevendo o problema de planejamento probabilístico, seja interpretando os dados devolvidos pelo simulador. Nosso sistema diminui essa dependência ao utilizar autômatos celulares para simular o tráfego e gerar informações que são então utilizadas para descrever o problema de planejamento probabilístico. Com isso podemos: (i) reduzir a necessidade de coleta de dados, que passam a ser gerados pelo simulador e (ii) produzir bons planos para o controle de semáforos de tempo fixo sem que seja necessária a intervenção de especialistas para a análise dos dados. Nos dois testes realizados a solução proposta pelo sistema diminuiu o tempo médio de percurso em 18:51% e 13:51%, respectivamente
Abstract: Traffic is an ever increasing problem, draining resources and aggravating pollution. In Sao Paulo, for instance, financial losses caused by traffic represent a sum of about R$33 billions a year. In this work we've developed a system that puts together the areas of Crowd Simulation and Probabilistic Planning to optimize fixed time traffic lights. Although both areas present good algorithms their use is limited by their reliance on specialists, whether to describe the probabilistic planning problem or to analyze the data produced by the simulations. Our approach contributes to minimize this dependence by using cellular automata simulations to generate the data that is used to describe the probabilistic planning problem. This allows us to: (i) reduce the amount of data collection, since the data is now generated by the simulator and (ii) produce good policies for fixed time traffic light control without the intervention of specialists to analyze the data. In the two tests performed the solution proposed by the system was able to reduce travel times by 18:51% and 13:51%, respectively
Mestrado
Ciência da Computação
Mestre em Ciência da Computação

Dans cette thèse, nous nous intéressons à la modélisation de mouvements de foules. Nous proposons un modèle microscopique basé sur la théorie des jeux. Chaque individu a une certaine vitesse souhaitée, celle qu'il adopterait en l'absence des autres. Une personne est influencée par certains de ses voisins, pratiquement ceux qu'elle voit devant elle. Une vitesse réelle est considérée comme possible si elle réalise un équilibre de Nash instantané: chaque individu fait son mieux par rapport à un objectif personnel (vitesse souhaitée), en tenant compte du comportement des voisins qui l'influencent. Nous abordons des questions relatives à la modélisation ainsi que les aspects théoriques du problème dans diverses situations, en particulier dans le cas où chaque individu est influencé par tous les autres, et le cas où les relations d'influence entre les individus présentent une structure hiérarchique. Un schéma numérique est développé pour résoudre le problème dans le second cas (modèle hiérarchique) et des simulations numériques sont proposées pour illustrer le comportement du modèle. Les résultats numériques sont confrontés avec des expériences réelles de mouvements de foules pour montrer la capacité du modèle à reproduire certains effets.Nous proposons une version macroscopique du modèle hiérarchique en utilisant les mêmes principes de modélisation au niveau macroscopique, et nous présentons une étude préliminaire des difficultés posées par cette approche.La dernière problématique qu'on aborde dans cette thèse est liée aux cadres flot gradient dans les espaces de Wasserstein aux niveaux continu et discret. Il est connu que l'équation de Fokker-Planck peut s'interpréter comme un flot gradient pour la distance de Wasserstein continue. Nous établissons un lien entre une discrétisation spatiale du type Volume Finis pour l'équation de Fokker-Planck sur une tesselation de Voronoï et les flots gradient sur le réseau sous-jacent, pour une distance de type Wasserstein récemment introduite sur l'espace de mesures portées par les sommets d'un réseaux
We are interested in the modeling of crowd motion. We propose a microscopic model based on game theoretic principles. Each individual is supposed to have a desired velocity, it is the one he would like to have in the absence of others. We consider that each individual is influenced by some of his neighbors, practically the ones that he sees. A possible actual velocity is an instantaneous Nash equilibrium: each individual does its best with respect to a personal objective (desired velocity), considering the behavior of the neighbors that influence him. We address theoretical and modeling issues in various situations, in particular when each individual is influenced by all the others, and in the case where the influence relations between individuals are hierarchical. We develop a numerical strategy to solve the problem in the second case (hierarchical model) and propose numerical simulations to illustrate the behavior of the model. We confront our numerical results with real experiments and prove the ability of the hierarchical model to reproduce some phenomena.We also propose to write a macroscopic counterpart of the hierarchical model by translating the same modeling principles to the macroscopic level and make the first steps towards writing such model.The last problem tackled in this thesis is related to gradient flow frameworks in the continuous and discrete Wasserstein spaces. It is known that the Fokker-Planck equation can be interpreted as a gradient flow for the continuous Wasserstein distance. We establish a link between some space discretization strategies of the Finite Volume type for the Fokker- Planck equation in general meshes (Voronoï tesselations) and gradient flows on the underlying networks of cells, in the framework of discrete Wasserstein-like distance on graphs recently introduced

Depuis l’article fondateur de Jordan, Kinderlehrer et Otto en 1998, il est bien connu qu’une large classe d’équations paraboliques peuvent être vues comme des flots de gradient dans l’espace de Wasserstein. Le but de cette thèse est d’étendre cette théorie à certaines équations et systèmes qui n’ont pas exactement une structure de flot de gradient. Les interactions étudiées sont de différentes natures. Le premier chapitre traite des systèmes avec des interactions non locales dans la dérive. Nous étudions ensuite des systèmes de diffusions croisées s’appliquant aux modèles de congestion pour plusieurs populations. Un autre modèle étudié est celui où le couplage se trouve dans le terme de réaction comme les systèmes proie-prédateur avec diffusion ou encore les modèles de croissance tumorale. Nous étudierons enfin des systèmes de type nouveau où l’interaction est donnée par un problème de transport multi-marges. Une grande partie de ces problèmes est illustrée de simulations numériques
Since 1998 and the seminal work of Jordan, Kinderlehrer and Otto, it is well known that a large class of parabolic equations can be seen as gradient flows in the Wasserstein space. This thesis is devoted to extensions of this theory to equations and systems which do not have exactly a gradient flow structure. We study different kind of couplings. First, we treat the case of nonlocal interactions in the drift. Then, we study cross diffusion systems which model congestion for several species. We are also interested in reaction-diffusion systems as diffusive prey-predator systems or tumor growth models. Finally, we introduce a new class of systems where the interaction is given by a multi-marginal transport problem. In many cases, we give numerical simulations to illustrate our theorical results

Trifft herabfallender Regen auf Waldflächen, so wird dieser Niederschlag umverteilt zu Interzeption, Stammabfluss, durchfallenden und abtropfenden Niederschlag. Besonders hohe Stammablaufmengen im Kronenzentrum und markante Abtropfstellen am Kronenrand einzelner Baumarten lassen sich zudem nur erklären, wenn am jeweiligen Messplatz unter Baumkronen eine weitere Niederschlagskomponente zugelassen wird, mit der laterale Wasserbewegungen innerhalb der expliziten Einzelbaumkrone beschrieben und bilanziert werden (lateraler [Zu- oder Ab-]Fluss). Ziel dieser Arbeit war es deshalb, der niederschlagsabhängigen und kleinräumigen Dynamik dieser Umverteilung im Wald am Beispiel der Baumarten Buche (Fagus sylvatica L.) und Fichte (Picea abies (L.) Karst.) durch Aufnahmen und Auswertungen in zwei Mischbeständen beider Baumarten nachzugehen. Wiederholende Niederschlagsaufzeichnungen einzelner Regenereignisse wurden hierzu in Relation zur jeweiligen Freilandniederschlagsmenge, zur Art der Messplatzüberschirmung im Kronendach und zur Distanz des Messplatzes zum ihn dominierenden Baum varianz- und regressionsanalytisch untersucht und um Ergebnisse aus Stammablaufanalysen und Streumodellen ergänzt. Auf dieser Basis konnten Kausalmodelle zur Schätzung des Unterkronenniederschlages mit hoher Güte hergeleitet werden, die im Detail ein baumartspezifisches räumliches Verhalten der einzelnen Niederschlagskomponenten beim Passieren der Baumkrone belegen. Neben den Besonderheiten, die sich bei Messungen in Bestandeslücken einstellen, ergeben sich so z.B. in der Vegetationsperiode je nach Messplatzposition und Freilandniederschlagsmenge Unterschiede im Niederschlag unter beiden Baumartenkronen von bis zu 35 % des Freilandniederschlages je Einzelereignis und zwar stets zugunsten der Buchenüberschirmung. Am Stammfuss von Buchen kann darüber hinaus zusätzliches Wasser durch den Stammabfluss eingetragen werden. Hierfür notwenige Wassermengen können plausibel aus dem nachgewiesenen lateralen Wasserabfluss im inneren Kronendrittel von Buchen gedeckt werden. Über ein räumlich konkretes Interzeptionsmodell, kombiniert mit Blattflächenschätzungen für Einzelbäume konkreter Dimension, konnte ein räumliches LAI-Modell für Buchen abgeleitet werden, dass höchste LAI-Werte im Kronenzentrum annimmt. Da der Bestandesniederschlag hinsichtlich Niederschlagsmenge, Bestockung, Belaubungszustand und zum räumlich konkreten Messplatz unter der einzelnen Baumkrone veränderlich ist, wird die gewissenhaftere Berücksichtigung dieser Einflussvariablen angeregt und der bisher häufig praktizierten pauschalen Aufteilung des Niederschlages in einzelne Niederschlagskomponenten auf Basis von Flächen- und Messphasendurchschnittswerten widersprochen
Forests redistribute the precipitation falling on their canopy into interception, stemflow, drip or direct throughfall. Extremely high amounts of stemflow in the centre of the crown and distinct drip points along the crown edge of certain tree species can only be explained by admitting an additional precipitation component at these measurement locations that describes and captures the lateral movement of water within the individual tree crown (lateral in- or outflow). The aim of this study was therefore to analyse these precipitation-dependent, small-scale dynamics of precipitation redistribution in forests using field-measurements from two mixed stands of European beech (Fagus sylvatica L.) and Norway spruce (Picea abies (L.) Karst.). Periodic measurements of individual precipitation events were examined in relation to the respective gross precipitation, the type of canopy above a plot and the distance of a plot to its dominant sheltering canopy tree using variance and regression analysis, and complemented with stemflow analyses and litterfall models. Using this dataset, causal models for the high-precision estimation of throughfall were derived, showing tree species-specific pathways of the individual precipitation components through the tree crown. Apart from the particularities of measurements in canopy gaps, differences in throughfall between spruce and beech during the vegetation period amount to up to 35 % of gross precipitation per event, in favour of the beech canopy and depending on plot location. At the stem base of beech trees additional water can reach the forest floor via stemflow. The amount of water required to generate this stemflow can plausibly be explained by means of the verified lateral water flow in the inner third of beech crowns. Using a spatially explicit interception model combined with LAI estimates for specific individual trees, a spatial LAI model was developed for beech, showing maximal LAI values in the crown centre. As the net forest precipitation is sensitive with respect to precipitation amount, stand type, foliage status and the spatially explicit plot location below an individual tree crown, this study recommends the consideration of these influential factors and contradicts the commonly practiced blanket partitioning of precipitation into individual components based on spatial and temporal averages

碩士
國立交通大學
資訊科學與工程研究所
105
Understanding the crowd flow and crowd density is crucial for smart city and urban planning. In this paper, we focus on the study of Mass rapid transit (MRT) that is playing an increasingly important role in many cities. The traditional way to estimate the crowd density and the crowd flow is by using smart card data. However, we can only know the number of passengers entering or exiting the station from smart card data. When and where the passengers change their MRT lines still remain unknown. Nowadays, each user has his/her own mobile phones and from the cellular data of mobile phone service providers, it is possible to know the users' transportation mode and the fine-grained crowd flows. As such, given a set of cellular data, we aim to estimate the crowd flow of MRT passengers and crowd density of stations as well as routes. To achieve these goals, we firstly propose an efficient and scalable approach to detect MRT trips with a pre-defined reference system. We take the cell tower properties, spatial and temporal factors into consideration in our approach. Then based on the detection result, we estimate the crowd flow and crowd density by grouping and counting the MRT trips. Extensive experiments are conducted to evaluate the detection and estimation approaches on a real dataset from Chunghwa Telecom, which is the largest telecommunication company in Taiwan. The results confirmed that our approaches are suitable for MRT trips detection, crowd flow and crowd density estimation. Finally, we provide case studies to present some applications and demonstrate the usefulness of our approaches.

A new prototype of subway station with overlaid passenger flow and information flow is suggested to be applied to the newly developed Line 8 in the City of Beijing, both to deal with the crowded passenger flow space and to provide a dynamically continuous city image to the passengers as a feedback of the collection of their route information.

We study the mean-field limit of a particle-based system modeling the behavior of many indistinguishable pedestrians as their number increases. The base model is a modified version of Helbing's social force model. In the mean-field limit, the time-dependent density of two-dimensional pedestrians satisfies a four-dimensional integro-differential Fokker-Planck equation. To approximate the solution of the Fokker-Planck equation we use a time-splitting approach and solve the diffusion part using a Crank-Nicholson method. The advection part is solved using a Lax-Wendroff-Leveque method or an upwind Backward Euler method depending on the advection speed. Moreover, we use multilevel Monte Carlo to estimate observables from the particle-based system. We discuss these numerical methods, and present numerical results showing the convergence of observables that were calculated using the particle-based model as the number of pedestrians increases to those calculated using the probability density function satisfying the Fokker-Planck equation.

碩士
中國文化大學
建築及都市設計學系
102
Abstract The Taipei subway transfer station is a busy interchange space which has multiple activities. It reflected the urban efficiency, safety indicators, traffic transport, humanity cultural exchange and the degree of happiness, but it also hidden potentially dangerous, including high density crowded, fire, spread of infectious diseases, natural disasters or terrorist attacks, therefore if the effective understanding of travelers dynamic, which can significantly reduce the space of potential danger. It's quite difficult to record crowd dynamic completely. The application of digital video storyboard can thoroughly recorded complete crowd dynamic. There are four theories in this study that are the environment behavior theory, self-driven particle, spatial statistical analysis and spatial information technology. Environment behavior is based on behavioral observation and through experimental design survey methodology to derive. Self-driven particles are like human having self-conscious, and being able to move on their own. Spatial statistical analysis is based on crowd of space statistical theory. Spatial information technology compared with the data collection and derived the calculation operation modes of work platforms, including overlay, internet, and temporal sequence. The process is first; through theoretical derivation and site scope survey, then preliminary investigation planning and behavioral observation. Secondary, Deployment of the camera zone codes, and then divides the investigation phase and investigation both officers paired and time control. Third, recording the sites, including digital imaging, dynamic video and writing records. To compiling the spatial statistical by using tables and preliminary survey data, and then outputs crowd survey spatial and temporal distribution, finally conclusions. In order to ensure the security of personal data, our research is using the lowest-resolution image to record, and to assess the effectiveness of its crowd survey. We survey of more than one thousand video records, including peak, off-peak and minor peak hours times the approximate spatial and temporal distribution of the crowd. To explored the preliminary of the potential risk of a spatial and temporal distribution, as the next stage of research applications subway transfer station performance assessment. Keywords：Crowd Behavior； Taipei Subway Transfer System； Spatial Statistical Analysis； Spatial Information Technology.

碩士
國立交通大學
資訊科學與工程研究所
107
Recently, predicting the crowd flows in a city has attracted a lot of attention because it is valuable to city transportation, management, and business. However, most of the previous works mainly focused on short-term prediction while long-term prediction is even more valuable for some important real-world applications, like traffic resources pre-allocations. As a result, this work aims to predict long-term crowd flows in a city by trajectory data or trip data obtained from mobile devices or GPS devices in vehicles. As a kind of time series data, crowd flows, including the inflow and the outflow, are influenced by both temporal dependencies and spatial dependencies. Therefore, we apply deep learning methods, such as convolution and residual structure, to extract short term, long short-term, and long term patterns. Besides, we also integrate the model with context data, including weather data and calendar information, to improve the performance. In addition, to enhance the robustness for long-term prediction and save computational resources, we adopt a multi-step architecture that can output predictions on multiple time steps simultaneously. To the best of our knowledge, this is the first work on long-term multi-step crowd flow prediction. The experiment results on two real-world datasets demonstrate that our framework outperforms the state-of-the-art methods on long-term crowd flow prediction. We also show that the multi-step architecture can enhance the execution efficiency without losing the accuracy performance.

Motion is an important cue in videos that captures the dynamics of moving objects. It helps in effective analysis of various event related tasks such as human action recognition, anomaly detection, tracking, crowd behavior analysis, traffic monitoring, etc. Generally, accurate motion information is computed using various optical flow estimation techniques. On the other hand, coarse motion information is readily available in the form of motion vectors in compressed videos. Utilizing these encoded motion vectors reduces the computational burden involved in flow estimation and enables rapid analysis of video streams. In this work, the focus is on analyzing motion patterns, retrieved from either motion vectors or optical flow, in order to do various event analysis tasks such as video classification, anomaly detection and crowd flow segmentation. In the first section, we utilize the motion vectors from H.264 compressed videos, a compression standard widely used due to its high compression ratio, to address the following problems. i) Video classification: This work proposes an approach to classify videos based on human action by capturing spatio-temporal motion pattern of the actions using Histogram of Oriented Motion Vector (HOMV) ii) Crowd flow segmentation: In this work, we have addressed the problem of flow segmentation of the dominant motion patterns of the crowds. The proposed approach combines multi-scale super-pixel segmentation of the motion vectors to obtain the final flow segmentation. iii) Anomaly detection: This problem is addressed by local modeling of usual behavior by capturing features such as magnitude and orientation of each moving object. In all the above approaches, the focus was to reduce computations while retaining comparable accuracy to pixel domain processing. In second section, we propose two approaches for anomaly detection using optical flow. The first approach uses spatio-temporal low level motion features and detects anomalies based on the reconstruction error of the sparse representation of the candidate feature over a dictionary of usual behavior features. The main contribution is in enhancing each local dictionary by applying an appropriate transformation on dictionaries of the neighboring regions. The other algorithm aims to improve the accuracy of anomaly localization through short local trajectories of super pixels belonging to moving objects. These trajectories capture both spatial as well as temporal information effectively. In contrast to compressed domain analysis, these pixel level approaches focus on improving the accuracy of detection with reasonable detection speed.

Trifft herabfallender Regen auf Waldflächen, so wird dieser Niederschlag umverteilt zu Interzeption, Stammabfluss, durchfallenden und abtropfenden Niederschlag. Besonders hohe Stammablaufmengen im Kronenzentrum und markante Abtropfstellen am Kronenrand einzelner Baumarten lassen sich zudem nur erklären, wenn am jeweiligen Messplatz unter Baumkronen eine weitere Niederschlagskomponente zugelassen wird, mit der laterale Wasserbewegungen innerhalb der expliziten Einzelbaumkrone beschrieben und bilanziert werden (lateraler [Zu- oder Ab-]Fluss). Ziel dieser Arbeit war es deshalb, der niederschlagsabhängigen und kleinräumigen Dynamik dieser Umverteilung im Wald am Beispiel der Baumarten Buche (Fagus sylvatica L.) und Fichte (Picea abies (L.) Karst.) durch Aufnahmen und Auswertungen in zwei Mischbeständen beider Baumarten nachzugehen. Wiederholende Niederschlagsaufzeichnungen einzelner Regenereignisse wurden hierzu in Relation zur jeweiligen Freilandniederschlagsmenge, zur Art der Messplatzüberschirmung im Kronendach und zur Distanz des Messplatzes zum ihn dominierenden Baum varianz- und regressionsanalytisch untersucht und um Ergebnisse aus Stammablaufanalysen und Streumodellen ergänzt. Auf dieser Basis konnten Kausalmodelle zur Schätzung des Unterkronenniederschlages mit hoher Güte hergeleitet werden, die im Detail ein baumartspezifisches räumliches Verhalten der einzelnen Niederschlagskomponenten beim Passieren der Baumkrone belegen. Neben den Besonderheiten, die sich bei Messungen in Bestandeslücken einstellen, ergeben sich so z.B. in der Vegetationsperiode je nach Messplatzposition und Freilandniederschlagsmenge Unterschiede im Niederschlag unter beiden Baumartenkronen von bis zu 35 % des Freilandniederschlages je Einzelereignis und zwar stets zugunsten der Buchenüberschirmung. Am Stammfuss von Buchen kann darüber hinaus zusätzliches Wasser durch den Stammabfluss eingetragen werden. Hierfür notwenige Wassermengen können plausibel aus dem nachgewiesenen lateralen Wasserabfluss im inneren Kronendrittel von Buchen gedeckt werden. Über ein räumlich konkretes Interzeptionsmodell, kombiniert mit Blattflächenschätzungen für Einzelbäume konkreter Dimension, konnte ein räumliches LAI-Modell für Buchen abgeleitet werden, dass höchste LAI-Werte im Kronenzentrum annimmt. Da der Bestandesniederschlag hinsichtlich Niederschlagsmenge, Bestockung, Belaubungszustand und zum räumlich konkreten Messplatz unter der einzelnen Baumkrone veränderlich ist, wird die gewissenhaftere Berücksichtigung dieser Einflussvariablen angeregt und der bisher häufig praktizierten pauschalen Aufteilung des Niederschlages in einzelne Niederschlagskomponenten auf Basis von Flächen- und Messphasendurchschnittswerten widersprochen.
Forests redistribute the precipitation falling on their canopy into interception, stemflow, drip or direct throughfall. Extremely high amounts of stemflow in the centre of the crown and distinct drip points along the crown edge of certain tree species can only be explained by admitting an additional precipitation component at these measurement locations that describes and captures the lateral movement of water within the individual tree crown (lateral in- or outflow). The aim of this study was therefore to analyse these precipitation-dependent, small-scale dynamics of precipitation redistribution in forests using field-measurements from two mixed stands of European beech (Fagus sylvatica L.) and Norway spruce (Picea abies (L.) Karst.). Periodic measurements of individual precipitation events were examined in relation to the respective gross precipitation, the type of canopy above a plot and the distance of a plot to its dominant sheltering canopy tree using variance and regression analysis, and complemented with stemflow analyses and litterfall models. Using this dataset, causal models for the high-precision estimation of throughfall were derived, showing tree species-specific pathways of the individual precipitation components through the tree crown. Apart from the particularities of measurements in canopy gaps, differences in throughfall between spruce and beech during the vegetation period amount to up to 35 % of gross precipitation per event, in favour of the beech canopy and depending on plot location. At the stem base of beech trees additional water can reach the forest floor via stemflow. The amount of water required to generate this stemflow can plausibly be explained by means of the verified lateral water flow in the inner third of beech crowns. Using a spatially explicit interception model combined with LAI estimates for specific individual trees, a spatial LAI model was developed for beech, showing maximal LAI values in the crown centre. As the net forest precipitation is sensitive with respect to precipitation amount, stand type, foliage status and the spatially explicit plot location below an individual tree crown, this study recommends the consideration of these influential factors and contradicts the commonly practiced blanket partitioning of precipitation into individual components based on spatial and temporal averages.