Shape complexity based on mutual information

Shape complexity has recently received attention from different fields, such as computer vision and psychology. In this paper, integral geometry and information theory tools are applied to quantify the shape complexity from two different perspectives: from the inside of the object, we evaluate its degree of structure or correlation between its surfaces (inner complexity), and from the outside, we compute its degree of interaction with the circumscribing sphere (outer complexity). Our shape complexity measures are based on the following two facts: uniformly distributed global lines crossing an object define a continuous information channel and the continuous mutual information of this channel is independent of the object discretisation and invariant to translations, rotations, and changes of scale. The measures introduced in this paper can be potentially used as shape descriptors for object recognition, image retrieval, object localisation, tumour analysis, and protein docking, among others

Classificació i reconeixement de vídeo

El principal objectiu d’aquest projecte és aconseguir classificar diferents vídeos d’esports segons la seva categoria. Els cercadors de text creen un vocabulari segons el significat de les diferents paraules per tal de poder identificar un document. En aquest projecte es va fer el mateix però mitjançant paraules visuals. Per exemple, es van intentar englobar com a una única paraula les diferents rodes que apareixien en els cotxes de rally. A partir de la freqüència amb què apareixien les paraules dels diferents grups dins d’una imatge vàrem crear histogrames de vocabulari que ens permetien tenir una descripció de la imatge. Per classificar un vídeo es van utilitzar els histogrames que descrivien els seus fotogrames. Com que cada histograma es podia considerar un vector de valors enters vàrem optar per utilitzar una màquina classificadora de vectors: una Support vector machine o SVM

Modelling stereoscopic vision systems for robotic applications

Aquesta tesi s'emmarca dins del projecte CICYT TAP 1999-0443-C05-01. L'objectiu d'aquest projecte és el disseny, implementació i avaluació de robots mòbils, amb un sistema de control distribuït, sistemes de sensorització i xarxa de comunicacions per realitzar tasques de vigilància. Els robots han de poder-se moure per un entorn reconeixent la posició i orientació dels diferents objectes que l'envolten. Aquesta informació ha de permetre al robot localitzar-se dins de l'entorn on es troba per poder-se moure evitant els possibles obstacles i dur a terme la tasca encomanada. El robot ha de generar un mapa dinàmic de l'entorn que serà utilitzat per localitzar la seva posició. L'objectiu principal d'aquest projecte és aconseguir que un robot explori i construeixi un mapa de l'entorn sense la necessitat de modificar el propi entorn. Aquesta tesi està enfocada en l'estudi de la geometria dels sistemes de visió estereoscòpics formats per dues càmeres amb l'objectiu d'obtenir informació geomètrica 3D de l'entorn d'un vehicle. Aquest objectiu tracta de l'estudi del modelatge i la calibració de càmeres i en la comprensió de la geometria epipolar. Aquesta geometria està continguda en el que s'anomena emph{matriu fonamental}. Cal realitzar un estudi del càlcul de la matriu fonamental d'un sistema estereoscòpic amb la finalitat de reduir el problema de la correspondència entre dos plans imatge. Un altre objectiu és estudiar els mètodes d'estimació del moviment basats en la geometria epipolar diferencial per tal de percebre el moviment del robot i obtenir-ne la posició. Els estudis de la geometria que envolta els sistemes de visió estereoscòpics ens permeten presentar un sistema de visió per computador muntat en un robot mòbil que navega en un entorn desconegut. El sistema fa que el robot sigui capaç de generar un mapa dinàmic de l'entorn a mesura que es desplaça i determinar quin ha estat el moviment del robot per tal de emph{localitzar-se} dins del mapa. La tesi presenta un estudi comparatiu dels mètodes de calibració de càmeres més utilitzats en les últimes dècades. Aquestes tècniques cobreixen un gran ventall dels mètodes de calibració clàssics. Aquest mètodes permeten estimar els paràmetres de la càmera a partir d'un conjunt de punts 3D i de les seves corresponents projeccions 2D en una imatge. Per tant, aquest estudi descriu un total de cinc tècniques de calibració diferents que inclouen la calibració implicita respecte l'explicita i calibració lineal respecte no lineal. Cal remarcar que s'ha fet un gran esforç en utilitzar la mateixa nomenclatura i s'ha estandaritzat la notació en totes les tècniques presentades. Aquesta és una de les dificultats principals a l'hora de poder comparar les tècniques de calibració ja què cada autor defineix diferents sistemes de coordenades i diferents conjunts de paràmetres. El lector és introduït a la calibració de càmeres amb la tècnica lineal i implícita proposada per Hall i amb la tècnica lineal i explicita proposada per Faugeras-Toscani. A continuació es passa a descriure el mètode a de Faugeras incloent el modelatge de la distorsió de les lents de forma radial. Seguidament es descriu el conegut mètode proposat per Tsai, i finalment es realitza una descripció detallada del mètode de calibració proposat per Weng. Tots els mètodes són comparats tant des del punt de vista de model de càmera utilitzat com de la precisió de la calibració. S'han implementat tots aquests mètodes i s'ha analitzat la precisió presentant resultats obtinguts tant utilitzant dades sintètiques com càmeres reals. Calibrant cada una de les càmeres del sistema estereoscòpic es poden establir un conjunt de restriccions geomètri ques entre les dues imatges. Aquestes relacions són el que s'anomena geometria epipolar i estan contingudes en la matriu fonamental. Coneixent la geometria epipolar es pot: simplificar el problema de la correspondència reduint l'espai de cerca a llarg d'una línia epipolar; estimar el moviment d'una càmera quan aquesta està muntada sobre un robot mòbil per realitzar tasques de seguiment o de navegació; reconstruir una escena per aplicacions d'inspecció, propotipatge o generació de motlles. La matriu fonamental s'estima a partir d'un conjunt de punts en una imatges i les seves correspondències en una segona imatge. La tesi presenta un estat de l'art de les tècniques d'estimació de la matriu fonamental. Comença pels mètode lineals com el dels set punts o el mètode dels vuit punts, passa pels mètodes iteratius com el mètode basat en el gradient o el CFNS, fins arribar las mètodes robustos com el M-Estimators, el LMedS o el RANSAC. En aquest treball es descriuen fins a 15 mètodes amb 19 implementacions diferents. Aquestes tècniques són comparades tant des del punt de vista algorísmic com des del punt de vista de la precisió que obtenen. Es presenten el resultats obtinguts tant amb imatges reals com amb imatges sintètiques amb diferents nivells de soroll i amb diferent quantitat de falses correspondències. Tradicionalment, l'estimació del moviment d'una càmera està basada en l'aplicació de la geometria epipolar entre cada dues imatges consecutives. No obstant el cas tradicional de la geometria epipolar té algunes limitacions en el cas d'una càmera situada en un robot mòbil. Les diferencies entre dues imatges consecutives són molt petites cosa que provoca inexactituds en el càlcul de matriu fonamental. A més cal resoldre el problema de la correspondència, aquest procés és molt costós en quant a temps de computació i no és gaire efectiu per aplicacions de temps real. En aquestes circumstàncies les tècniques d'estimació del moviment d'una càmera solen basar-se en el flux òptic i en la geometria epipolar diferencial. En la tesi es realitza un recull de totes aquestes tècniques degudament classificades. Aquests mètodes són descrits unificant la notació emprada i es remarquen les semblances i les diferencies entre el cas discret i el cas diferencial de la geometria epipolar. Per tal de poder aplicar aquests mètodes a l'estimació de moviment d'un robot mòbil, aquest mètodes generals que estimen el moviment d'una càmera amb sis graus de llibertat, han estat adaptats al cas d'un robot mòbil que es desplaça en una superfície plana. Es presenten els resultats obtinguts tant amb el mètodes generals de sis graus de llibertat com amb els adaptats a un robot mòbil utilitzant dades sintètiques i seqüències d'imatges reals. Aquest tesi finalitza amb una proposta de sistema de localització i de construcció d'un mapa fent servir un sistema estereoscòpic situat en un robot mòbil. Diverses aplicacions de robòtica mòbil requereixen d'un sistema de localització amb l'objectiu de facilitar la navegació del vehicle i l'execució del les trajectòries planificades. La localització es sempre relativa al mapa de l'entorn on el robot s'està movent. La construcció de mapes en un entorn desconegut és una tasca important a realitzar per les futures generacions de robots mòbils. El sistema que es presenta realitza la localització i construeix el mapa de l'entorn de forma simultània. A la tesi es descriu el robot mòbil GRILL, que ha estat la plataforma de treball emprada per aquesta aplicació, amb el sistema de visió estereoscòpic que s'ha dissenyat i s'ha muntat en el robot. També es descriu tots el processos que intervenen en el sistema de localització i construcció del mapa. La implementació d'aquest processos ha estat possible gràcies als estudis realitzats i presentats prèviament (calibració de càmeres, estimació de la matriu fonamental, i estimació del moviment) sense els quals no s'hauria pogut plantejar aquest sistema. Finalment es presenten els mapes en diverses trajectòries realitzades pel robot GRILL en el laboratori. Les principals contribucions d'aquest treball són: ·Un estat de l'art sobre mètodes de calibració de càmeres. El mètodes són comparats tan des del punt de vista del model de càmera utilitzat com de la precisió dels mètodes. ·Un estudi dels mètodes d'estimació de la matriu fonamental. Totes les tècniques estudiades són classificades i descrites des d'un punt de vista algorísmic. ·Un recull de les tècniques d'estimació del moviment d'una càmera centrat en el mètodes basat en la geometria epipolar diferencial. Aquestes tècniques han estat adaptades per tal d'estimar el moviment d'un robot mòbil. ·Una aplicació de robòtica mòbil per tal de construir un mapa dinàmic de l'entorn i localitzar-se per mitja d'un sistema estereoscòpic. L'aplicació presentada es descriu tant des del punt de vista del maquinari com del programari que s'ha dissenyat i implementat.

Biologically motivated model for object detection and identification in real-world scenes

The classical computer vision methods can only weakly emulate some of the multi-level parallelisms in signal processing and information sharing that takes place in different parts of the primates’ visual system thus enabling it to accomplish many diverse functions of visual perception. One of the main functions of the primates’ vision is to detect and recognise objects in natural scenes despite all the linear and non-linear variations of the objects and their environment. The superior performance of the primates’ visual system compared to what machine vision systems have been able to achieve to date, motivates scientists and researchers to further explore this area in pursuit of more efficient vision systems inspired by natural models. In this paper building blocks for a hierarchical efficient object recognition model are proposed. Incorporating the attention-based processing would lead to a system that will process the visual data in a non-linear way focusing only on the regions of interest and hence reducing the time to achieve real-time performance. Further, it is suggested to modify the visual cortex model for recognizing objects by adding non-linearities in the ventral path consistent with earlier discoveries as reported by researchers in the neuro-physiology of vision.

Marr's vision: twenty-five years on

It is twenty-five years since the posthumous publication of David Marr's book Vision [1]. Only 35 years old when he died, Man, had already dramatically influenced vision research. His book, and the series of papers that preceded it, have had a lasting impact on the way that researchers approach human and computer vision.

Empirical impact of cache effects for vision system design

Computer vision applications generally split their problem into multiple simpler tasks. Likewise research often combines algorithms into systems for evaluation purposes. Frameworks for modular vision provide interfaces and mechanisms for algorithm combination and network transparency. However, these don’t provide interfaces efficiently utilising the slow memory in modern PCs. We investigate quantitatively how system performance varies with different patterns of memory usage by the framework for an example vision system.

FACSGen: a tool to synthesize emotional facial expressions through systematic manipulation of facial action units

To investigate the perception of emotional facial expressions, researchers rely on shared sets of photos or videos, most often generated by actor portrayals. The drawback of such standardized material is a lack of flexibility and controllability, as it does not allow the systematic parametric manipulation of specific features of facial expressions on the one hand, and of more general properties of the facial identity (age, ethnicity, gender) on the other. To remedy this problem, we developed FACSGen: a novel tool that allows the creation of realistic synthetic 3D facial stimuli, both static and dynamic, based on the Facial Action Coding System. FACSGen provides researchers with total control over facial action units, and corresponding informational cues in 3D synthetic faces. We present four studies validating both the software and the general methodology of systematically generating controlled facial expression patterns for stimulus presentation.

Learning enhanced 3D models for vehicle tracking

This paper presents an enhanced hypothesis verification strategy for 3D object recognition. A new learning methodology is presented which integrates the traditional dichotomic object-centred and appearance-based representations in computer vision giving improved hypothesis verification under iconic matching. The "appearance" of a 3D object is learnt using an eigenspace representation obtained as it is tracked through a scene. The feature representation implicitly models the background and the objects observed enabling the segmentation of the objects from the background. The method is shown to enhance model-based tracking, particularly in the presence of clutter and occlusion, and to provide a basis for identification. The unified approach is discussed in the context of the traffic surveillance domain. The approach is demonstrated on real-world image sequences and compared to previous (edge-based) iconic evaluation techniques.

Benchmark datasets for detection and tracking

There is a rising demand for the quantitative performance evaluation of automated video surveillance. To advance research in this area, it is essential that comparisons in detection and tracking approaches may be drawn and improvements in existing methods can be measured. There are a number of challenges related to the proper evaluation of motion segmentation, tracking, event recognition, and other components of a video surveillance system that are unique to the video surveillance community. These include the volume of data that must be evaluated, the difficulty in obtaining ground truth data, the definition of appropriate metrics, and achieving meaningful comparison of diverse systems. This chapter provides descriptions of useful benchmark datasets and their availability to the computer vision community. It outlines some ground truth and evaluation techniques, and provides links to useful resources. It concludes by discussing the future direction for benchmark datasets and their associated processes.

Empathy and emotion recognition in people with autism, first-degree relatives, and controls

Empathy is the lens through which we view others' emotion expressions, and respond to them. In this study, empathy and facial emotion recognition were investigated in adults with autism spectrum conditions (ASC; N=314), parents of a child with ASC (N=297) and IQ-matched controls (N=184). Participants completed a self-report measure of empathy (the Empathy Quotient [EQ]) and a modified version of the Karolinska Directed Emotional Faces Task (KDEF) using an online test interface. Results showed that mean scores on the EQ were significantly lower in fathers (p<0.05) but not mothers (p>0.05) of children with ASC compared to controls, whilst both males and females with ASC obtained significantly lower EQ scores (p<0.001) than controls. On the KDEF, statistical analyses revealed poorer overall performance by adults with ASC (p<0.001) compared to the control group. When the 6 distinct basic emotions were analysed separately, the ASC group showed impaired performance across five out of six expressions (happy, sad, angry, afraid and disgusted). Parents of a child with ASC were not significantly worse than controls at recognising any of the basic emotions, after controlling for age and non-verbal IQ (all p>0.05). Finally, results indicated significant differences between males and females with ASC for emotion recognition performance (p<0.05) but not for self-reported empathy (p>0.05). These findings suggest that self-reported empathy deficits in fathers of autistic probands are part of the 'broader autism phenotype'. This study also reports new findings of sex differences amongst people with ASC in emotion recognition, as well as replicating previous work demonstrating empathy difficulties in adults with ASC. The use of empathy measures as quantitative endophenotypes for ASC is discussed.

A survey of human motion analysis using depth imagery

Analysis of human behaviour through visual information has been a highly active research topic in the computer vision community. This was previously achieved via images from a conventional camera, but recently depth sensors have made a new type of data available. This survey starts by explaining the advantages of depth imagery, then describes the new sensors that are available to obtain it. In particular, the Microsoft Kinect has made high-resolution real-time depth cheaply available. The main published research on the use of depth imagery for analysing human activity is reviewed. Much of the existing work focuses on body part detection and pose estimation. A growing research area addresses the recognition of human actions. The publicly available datasets that include depth imagery are listed, as are the software libraries that can acquire it from a sensor. This survey concludes by summarising the current state of work on this topic, and pointing out promising future research directions.

On the 'special' status of emotional faces... Comment on Yang, Hong, and Blake (2010)

A wealth of literature suggests that emotional faces are given special status as visual objects: Cognitive models suggest that emotional stimuli, particularly threat-relevant facial expressions such as fear and anger, are prioritized in visual processing and may be identified by a subcortical “quick and dirty” pathway in the absence of awareness (Tamietto & de Gelder, 2010). Both neuroimaging studies (Williams, Morris, McGlone, Abbott, & Mattingley, 2004) and backward masking studies (Whalen, Rauch, Etcoff, McInerney, & Lee, 1998) have supported the notion of emotion processing without awareness. Recently, our own group (Adams, Gray, Garner, & Graf, 2010) showed adaptation to emotional faces that were rendered invisible using a variant of binocular rivalry: continual flash suppression (CFS, Tsuchiya & Koch, 2005). Here we (i) respond to Yang, Hong, and Blake's (2010) criticisms of our adaptation paper and (ii) provide a unified account of adaptation to facial expression, identity, and gender, under conditions of unawareness

A unified approach to the recognition of complex actions from sequences of zone-crossings

We present a method for the recognition of complex actions. Our method combines automatic learning of simple actions and manual definition of complex actions in a single grammar. Contrary to the general trend in complex action recognition that consists in dividing recognition into two stages, our method performs recognition of simple and complex actions in a unified way. This is performed by encoding simple action HMMs within the stochastic grammar that models complex actions. This unified approach enables a more effective influence of the higher activity layers into the recognition of simple actions which leads to a substantial improvement in the classification of complex actions. We consider the recognition of complex actions based on person transits between areas in the scene. As input, our method receives crossings of tracks along a set of zones which are derived using unsupervised learning of the movement patterns of the objects in the scene. We evaluate our method on a large dataset showing normal, suspicious and threat behaviour on a parking lot. Experiments show an improvement of ~ 30% in the recognition of both high-level scenarios and their composing simple actions with respect to a two-stage approach. Experiments with synthetic noise simulating the most common tracking failures show that our method only experiences a limited decrease in performance when moderate amounts of noise are added.

MACHINE VISION FOR AUTOMATICVISUAL INSPECTION OF WOODENRAILWAY SLEEPERS USING UNSUPERVISED NEURAL NETWORKS

The motivation for this thesis work is the need for improving reliability of equipment and quality of service to railway passengers as well as a requirement for cost-effective and efficient condition maintenance management for rail transportation. This thesis work develops a fusion of various machine vision analysis methods to achieve high performance in automation of wooden rail track inspection.The condition monitoring in rail transport is done manually by a human operator where people rely on inference systems and assumptions to develop conclusions. The use of conditional monitoring allows maintenance to be scheduled, or other actions to be taken to avoid the consequences of failure, before the failure occurs. Manual or automated condition monitoring of materials in fields of public transportation like railway, aerial navigation, traffic safety, etc, where safety is of prior importance needs non-destructive testing (NDT).In general, wooden railway sleeper inspection is done manually by a human operator, by moving along the rail sleeper and gathering information by visual and sound analysis for examining the presence of cracks. Human inspectors working on lines visually inspect wooden rails to judge the quality of rail sleeper. In this project work the machine vision system is developed based on the manual visual analysis system, which uses digital cameras and image processing software to perform similar manual inspections. As the manual inspection requires much effort and is expected to be error prone sometimes and also appears difficult to discriminate even for a human operator by the frequent changes in inspected material. The machine vision system developed classifies the condition of material by examining individual pixels of images, processing them and attempting to develop conclusions with the assistance of knowledge bases and features.A pattern recognition approach is developed based on the methodological knowledge from manual procedure. The pattern recognition approach for this thesis work was developed and achieved by a non destructive testing method to identify the flaws in manually done condition monitoring of sleepers.In this method, a test vehicle is designed to capture sleeper images similar to visual inspection by human operator and the raw data for pattern recognition approach is provided from the captured images of the wooden sleepers. The data from the NDT method were further processed and appropriate features were extracted.The collection of data by the NDT method is to achieve high accuracy in reliable classification results. A key idea is to use the non supervised classifier based on the features extracted from the method to discriminate the condition of wooden sleepers in to either good or bad. Self organising map is used as classifier for the wooden sleeper classification.In order to achieve greater integration, the data collected by the machine vision system was made to interface with one another by a strategy called fusion. Data fusion was looked in at two different levels namely sensor-level fusion, feature- level fusion. As the goal was to reduce the accuracy of the human error on the rail sleeper classification as good or bad the results obtained by the feature-level fusion compared to that of the results of actual classification were satisfactory.

Graphical models and point set matching

Point pattern matching in Euclidean Spaces is one of the fundamental problems in Pattern Recognition, having applications ranging from Computer Vision to Computational Chemistry. Whenever two complex patterns are encoded by two sets of points identifying their key features, their comparison can be seen as a point pattern matching problem. This work proposes a single approach to both exact and inexact point set matching in Euclidean Spaces of arbitrary dimension. In the case of exact matching, it is assured to find an optimal solution. For inexact matching (when noise is involved), experimental results confirm the validity of the approach. We start by regarding point pattern matching as a weighted graph matching problem. We then formulate the weighted graph matching problem as one of Bayesian inference in a probabilistic graphical model. By exploiting the existence of fundamental constraints in patterns embedded in Euclidean Spaces, we prove that for exact point set matching a simple graphical model is equivalent to the full model. It is possible to show that exact probabilistic inference in this simple model has polynomial time complexity with respect to the number of elements in the patterns to be matched. This gives rise to a technique that for exact matching provably finds a global optimum in polynomial time for any dimensionality of the underlying Euclidean Space. Computational experiments comparing this technique with well-known probabilistic relaxation labeling show significant performance improvement for inexact matching. The proposed approach is significantly more robust under augmentation of the sizes of the involved patterns. In the absence of noise, the results are always perfect.