Qian / Wu / Xu Intelligent Surveillance Systems

1. Auflage 2011
ISBN: 978-94-007-1137-2
Verlag: Springer Netherlands
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, Band 51, 168 Seiten

Reihe: Intelligent Systems, Control and Automation: Science and Engineering

Surveillance systems have become increasingly popular. Full involvement of human operators has led to shortcomings, e.g. high labor cost, limited capability for multiple screens, inconsistency in long-duration, etc. Intelligent surveillance systems (ISSs) can supplement or even replace traditional ones. In ISSs, computer vision, pattern recognition, and artificial intelligence technologies are used to identify abnormal behaviours in videos. They present the development of real-time behaviour-based intelligent surveillance systems. The book focuses on the detection of individual abnormal behaviour based on learning and the analysis of dangerous crowd behaviour based on texture and optical flow. Practical systems include a real-time face classification and counting system, a surveillance robot system that utilizes video and audio information for intelligent interaction, and a robust person counting system for crowded environments.

Autoren/Hrsg.

Qian, Huihuan

Wu, Xinyu

Xu, Yangsheng

Inhaltsverzeichnis

1;Preface;7
2;Contents;8
3;List of Figures;11
4;List of Tables;16
5;Chapter 1Introduction;17
5.1;1.1 Background;17
5.2;1.2 Existing Surveillance Systems;18
5.3;1.3 Book Contents;19
5.4;1.4 Conclusion;21
6;Chapter 2 Background/Foreground Detection;22
6.1;2.1 Introduction;22
6.2;2.2 Pattern Classification Method;22
6.2.1;2.2.1 Overview of Background Update Methods ;22
6.2.1.1;2.2.1.1 Multi-Frame Average Method;23
6.2.1.2;2.2.1.2 Selection Method ;23
6.2.1.3;2.2.1.3 Selection-Average Method ;24
6.2.1.4;2.2.1.4 Kalman Filter-based Adaptive Background Update Method;24
6.2.1.5;2.2.1.5 Another Adaptive Background Update Method;24
6.2.1.6;2.2.1.6 Current Applications of Background Update Methods;25
6.2.2;2.2.2 Pattern Classification-based Adaptive Background Update Method;26
6.3;2.3 Frame Differencing Method;31
6.4;2.4 Optical Flow Method;35
6.5;2.5 Conclusion;36
7;Chapter 3 Segmentation and Tracking;37
7.1;3.1 Introduction;37
7.2;3.2 Segmentation ;37
7.3;3.3 Tracking;42
7.3.1;3.3.1 Hybrid Tracking Method ;42
7.3.1.1;3.3.1.1 Distance Tracking ;42
7.3.1.2;3.3.1.2 Color Tracking ;43
7.3.1.3;3.3.1.3 Fusion of the Two Tracking Approaches;45
7.3.1.4;3.3.1.4 Experimental Study;45
7.3.2;3.3.2 Particle Filter-based Tracking Method ;45
7.3.2.1;3.3.2.1 Target Model Update;48
7.3.3;3.3.3 Local Binary Pattern-based Tracking Method ;49
7.3.3.1;3.3.3.1 Multiple Target Tracking;52
7.3.3.2;3.3.3.2 Kalman Filter;52
7.3.3.3;3.3.3.3 LBP Histogram Distance;54
7.3.3.4;3.3.3.4 Blob Classification ;54
7.3.3.5;3.3.3.5 Experiment and Discussion;55
7.4;3.4 Conclusion;57
8;Chapter 4 Behavior Analysis of Individuals;58
8.1;4.1 Introduction;58
8.2;4.2 Learning-based Behavior Analysis ;58
8.2.1;4.2.1 Contour-based Feature Analysis ;58
8.2.1.1;4.2.1.1 Preprocessing;58
8.2.1.2;4.2.1.2 Supervised PCA for Feature Generation ;59
8.2.1.3;4.2.1.3 SVM classifiers;60
8.2.1.4;4.2.1.4 Experiments;61
8.2.2;4.2.2 Motion-based Feature Analysis ;62
8.2.2.1;4.2.2.1 Mean Shift-based Motion Feature Searching ;62
8.2.2.2;4.2.2.2 Motion History Image-based Analysis;64
8.2.2.3;4.2.2.3 Frame Work Analysis;65
8.2.2.4;4.2.2.4 SVM-based Learning;66
8.2.2.5;4.2.2.5 Recognition using a Bayesian Network;66
8.2.2.6;4.2.2.6 Experiments;67
8.3;4.3 Rule-based Behavior Analysis ;68
8.4;4.4 Application: Household Surveillance Robot ;69
8.4.1;4.4.1 System Implementation;73
8.4.2;4.4.2 Combined Surveillance with Video and Audio ;74
8.4.2.1;4.4.2.1 MFCC Feature Extraction ;75
8.4.2.2;4.4.2.2 Support Vector Machine ;77
8.4.3;4.4.3 Experimental Results;78
8.5;4.5 Conclusion;81
9;Chapter 5 Facial Analysis of Individuals;83
9.1;5.1 Feature Extraction ;85
9.1.1;5.1.1 Supervised PCA for Feature Generation;85
9.1.2;5.1.2 ICA-based Feature Extraction;87
9.2;5.2 Fusion of SVM Classifiers;88
9.3;5.3 System and Experiments;90
9.3.1;5.3.1 Implementation;91
9.3.2;5.3.2 Experiment Result;92
9.4;5.4 Conclusion;92
10;Chapter 6 Behavior Analysis of Human Groups;93
10.1;6.1 Introduction;93
10.2;6.2 Agent Tracking and Status Analysis ;94
10.3;6.3 Group Analysis ;95
10.3.1;6.3.1 Queuing ;97
10.3.2;6.3.2 Gathering and Dispersing ;99
10.4;6.4 Experiments;100
10.4.1;6.4.1 Multi-Agent Queuing;102
10.4.2;6.4.2 Gathering and Dispersing;102
10.5;6.5 Conclusion;103
11;Chapter 7 Static Analysis of Crowds: Human Counting and Distribution;104
11.1;7.1 Blob-based Human Counting and Distribution ;104
11.1.1;7.1.1 Overview;105
11.1.2;7.1.2 Preprocessing;107
11.1.3;7.1.3 Input Selection;107
11.1.4;7.1.4 Blob Learning ;109
11.1.5;7.1.5 Experiments;110
11.1.6;7.1.6 Conclusion;112
11.2;7.2 Feature-based Human Counting and Distribution ;112
11.2.1;7.2.1 Overview;113
11.2.2;7.2.2 Initial Calibration;115
11.2.2.1;7.2.2.1 Multiresolution Density Cells with a Perspective Projection Model ;115
11.2.2.2;7.2.2.2 Normalization of Density Cells ;119
11.2.3;7.2.3 Density Estimation ;119
11.2.3.1;7.2.3.1 Feature Extraction;120
11.2.3.2;7.2.3.2 Searching for the Characteristic Scale ;121
11.2.3.3;7.2.3.3 System Training;123
11.2.4;7.2.4 Detection of an Abnormal Density Distribution ;123
11.2.4.1;7.2.4.1 Training Data Created by Computer Simulation;123
11.2.4.2;7.2.4.2 System Training and Testing;124
11.2.5;7.2.5 Experiment Results;125
11.2.6;7.2.6 Conclusion;128
12;Chapter 8 Dynamic Analysis of Crowd Behavior;129
12.1;8.1 Behavior Analysis of Individuals in Crowds ;129
12.2;8.2 Energy-based Behavior Analysis of Groups in Crowds ;130
12.2.1;8.2.1 First Video Energy ;133
12.2.1.1;8.2.1.1 Definition of First Video Energy;133
12.2.1.2;8.2.1.2 Quartation Algorithm ;135
12.2.2;8.2.2 Second Video Energy ;137
12.2.2.1;8.2.2.1 Motion Feature;137
12.2.2.2;8.2.2.2 Definition of Second Video Energy;139
12.2.3;8.2.3 Third Video Energy;141
12.2.3.1;8.2.3.1 Definition of Third Video Energy;141
12.2.3.2;8.2.3.2 Angle Field Analysis ;142
12.2.3.3;8.2.3.3 Weighted Coefficients Design;142
12.2.4;8.2.4 Experiment using a Metro Surveillance System ;143
12.2.4.1;8.2.4.1 Description of Abnormality ;144
12.2.4.2;8.2.4.2 Wavelet Analysis ;144
12.2.4.3;8.2.4.3 Comparison of Two Kinds of Video Energy;146
12.2.5;8.2.5 Experiment Using an ATM Surveillance System ;147
12.2.5.1;8.2.5.1 Sensitive Area Monitoring Subsystem;148
12.2.5.2;8.2.5.2 Aggressive Behaviors Detection Subsystem;148
12.2.5.3;8.2.5.3 Logical Decision-Making Subsystem;149
12.2.5.4;8.2.5.4 Experiments;149
12.3;8.3 RANSAC-based Behavior Analysis of Groups in Crowds ;156
12.3.1;8.3.1 Random Sample Consensus (RANSAC) ;156
12.3.2;8.3.2 Estimation of Crowd Flow Direction ;158
12.3.3;8.3.3 Definition of a Group in a Crowd (Crowd Group) ;160
12.3.4;8.3.4 Experiment and Discussion;162
13;References;165
14;Index;175