Tracking with Particle Filter for High-dimensional Observation and State Spaces

This title concerns the use of a particle filter framework to track objects defined in high-dimensional state-spaces using high-dimensional observation spaces. Current tracking applications require us to consider complex models for objects (articulated objects, multiple objects, multiple fragments, etc.) as well as multiple kinds of information (multiple cameras, multiple modalities, etc.). This book presents some recent research that considers the main bottleneck of particle filtering frameworks (high dimensional state spaces) for tracking in such difficult conditions.

Table of Contents:
NOTATIONS ix INTRODUCTION xi CHAPTER 1. VISUAL TRACKING BY PARTICLE FILTERING 1 1.1. Introduction 1 1.2. Theoretical models 2 1.2.1. Recursive Bayesian filtering 2 1.2.2. Sequential Monte-Carlo methods 4 1.2.3. Application to visual tracking 8 1.3. Limits and challenges 18 1.4. Scientific position 22 1.5. Managing large sizes in particle filtering 22 1.6. Conclusion 26 CHAPTER 2. DATA REPRESENTATION MODELS 29 2.1. Introduction 29 2.2. Computation of the likelihood function    30 2.2.1. Exploitation of the spatial redundancy   31 2.2.2. Exploitation of the temporal redundancy 42 2.3. Representation of complex information    50 2.3.1. Representation of observations for movement detection, appearances and disappearances 50 2.3.2. Representation of deformations 53 2.3.3. Multifeature representation 56 2.4. Conclusion 75 CHAPTER 3. TRACKING MODELS THAT FOCUS ON THE STATE SPACE 79 3.1. Introduction 79 3.2. Data association methods for multi-object tracking  80 3.2.1. Particle filter with adaptive classification  84 3.2.2. Energetic filter for data association    87 3.3. Introducing fuzzy information into the particle filter 95 3.3.1. Fuzzy representation   96 3.3.2. Fuzzy spatial relations  98 3.3.3. Integration of fuzzy spatial relations into the particle filter   99 3.4. Conjoint estimation of dynamic and static parameters 114 3.5. Conclusion 119 CHAPTER 4. MODELS OF TRACKING BY DECOMPOSITION OF THE STATE SPACE  123 4.1. Introduction 123 4.2. Ranked partitioned sampling       126 4.3. Weighted partitioning with permutation of sub-particles 133 4.3.1. Permutation of sub-samples 135 4.3.2. Decrease the number of resamplings 138 4.3.3. General algorithm and results 138 4.4. Combinatorial resampling 142 4.5. Conclusion 149 CHAPTER 5. RESEARCH PERSPECTIVES IN TRACKING AND MANAGING LARGE SPACES 151 5.1. Tracking for behavioral analysis: toward finer tracking of the “future” and the “now”  153 5.2. Tracking for event detection: toward a top-down model 156 5.3. Tracking to measure social interactions 159 BIBLIOGRAPHY 163 INDEX   197