Contextual analysis of videos / Myo Thida [and three others].

Part of: Synthesis digital library of engineering and computer science.

Series from website.

Includes bibliographical references : (pages 77-91).

1. Introduction -- 1.1 Aims and objectives -- 1.2 Challenges -- 1.3 Nomenclature -- 1.4 Contributions -- 1.5 Organisation --

2. Literature review -- 2.1 Overview -- 2.2 Tracking multiple targets -- 2.2.1 Tracking multiple targets using particle filter -- 2.2.2 Tracking multiple targets using additional cues -- 2.2.3 Multiple-camera tracking -- 2.3 Analysis of crowd behaviour -- 2.3.1 Abnormality detection using micro-observation -- 2.3.2 Abnormality detection using macro-observation -- 2.3.3 Event detection -- 2.3.4 Graph-based and manifold learning algorithms -- 2.4 Summary --

3. Tracking multiple targets using particle swarm optimisation -- 3.1 Introduction -- 3.2 Literature review on particle swarm optimisation -- 3.3 Standard particle swarm optimisation -- 3.3.1 Convergence criteria -- 3.3.2 Pseudo-code -- 3.4 A modified PSO with interactive swarms -- 3.4.1 Particle and swarm diversification -- 3.4.2 Swarm optimisation -- 3.4.3 Swarm initialisation and termination -- 3.4.4 Algorithm summary -- 3.5 Experiments -- 3.5.1 Tracking fixed and known number of targets -- 3.5.2 Tracking unknown and varying number of targets -- 3.5.3 Performance evaluation -- 3.6 Summary --

4. Abnormality detection in crowded scenes -- 4.1 Introduction -- 4.2 Global abnormality detection -- 4.2.1 Frame-based video representation -- 4.2.2 Spatio-temporal Laplacian Eigenmaps -- 4.2.3 Analysing video manifolds in temporal domain -- 4.2.4 Experimental results -- 4.3 Local abnormality detection -- 4.3.1 Representation of local motion -- 4.3.2 Temporally constrained Laplacian Eigenmaps -- 4.3.3 Representation of regular motion pattern -- 4.3.4 Abnormality detection -- 4.3.5 Abnormality localisation -- 4.3.6 Experimental results -- 4.4 Summary --

5. Conclusion -- 5.1 Future directions -- Bibliography -- Authors' biographies.

Video context analysis is an active and vibrant research area, which provides means for extracting, analyzing and understanding behavior of a single target and multiple targets. Over the last few decades, computer vision researchers have been working to improve the accuracy and robustness of algorithms to analyze the context of a video automatically. In general, the research work in this area can be categorized into three major topics: 1) counting number of people in the scene 2) tracking individuals in a crowd and 3) understanding behavior of a single target or multiple targets in the scene. This book focuses on tracking individual targets and detecting abnormal behavior of a crowd in a complex scene