Robust Optical Flow Estimation in Outdoor Environments

About the Authors xiii
Preface xv
Acknowledgments xvii
1 Introduction 1
1.1 Introduction 1
1.2 Background and Significance 4
1.3 Optical Flow Estimation Applications 7
1.4 Current Research Trends and Problems 10
1.5 Structure of the Book 13
Examples with Solutions 15
References 17
2 Fundamentals of Optical Flow 21
2.1 Classical Optical Flow Estimation Models 21
2.2 Variational Optical Flow Estimation Methods 22
2.3 Total Variation (TV-L1 ) Optical Flow Estimation 24
2.4 Noise and Uncertainty in Outdoor Environments 27
2.5 Techniques for Robust Motion Estimation Under Noise and Uncertainty 31
2.6 Uncertainty-Aware Motion Estimation Methods 32
2.7 Real-World Applications of Noise and Uncertainty in Motion Estimation 33
2.8 Handling Motion in Low-Light and Foggy Conditions 35
2.9 Optical Flow for Rain, Snow, and Other Weather Conditions 36
2.10 Deep Learning-based Optical Flow Estimation 37
2.11 Optical Flow with Recurrent Neural Networks 38
2.12 Self-Supervised and Unsupervised Learning Approaches 40
2.13 Optical Flow in 3D Vision 41
2.14 Optical Flow Estimation-based Motion Segmentation 41
2.15 Evaluation Measures 44
2.16 Datasets 45
2.17 Conclusion 46
Examples with Solutions 47
References 53
3 Robust NLTV-L 1 Optical Flow Estimation Under Adverse Outdoor Conditions 59
3.1 Introduction 59
3.2 Non-local TV-L1 for Optical Flow Estimation 61
3.3 RNLTV-L1 Optical Flow Estimation Method 63
3.4 Motion Pattern-based Segmentation 68
3.5 Incorporating Deep Learning for Motion Segmentation 70
3.6 Fusion of Appearance and Motion-based Features 71
3.7 Experiments and Analysis 72
3.8 Results 75
3.9 Conclusion 82
Examples with Solutions 87
References 93
4 Robust Optical Flow Estimation via Edge-preserving Filter 97
4.1 Introduction 97
4.2 Robust WGF for the Optical Flow Estimation 99
4.3 Robust NLTV-L1 +WGF for Optical Flow Estimation 101
4.4 Experiments and Analysis 103
4.5 Results 105
4.6 Conclusion 110
Examples with Solutions 111
References 115
5 Optical Flow Estimation via Weighted Guided Filter with Nonlocal Steering Kernel 119
5.1 Introduction 119
5.2 The WGF with a Nonlocal Steering Kernel for Optical Flow Estimation 120
5.3 Optimization of the Optical Flow Estimation 123
5.4 Experiments and Analysis 125
5.5 Conclusion 130
Examples with Solutions 131
References 141
6 Advance Deep Learning Approaches for Optical Flow Estimation 145
6.1 Introduction 145
6.2 Convolutional Neural Networks for Optical Flow Estimation 145
6.3 Unsupervised and Self-supervised Learning Approaches 147
6.4 Conclusion 148
Examples with Solutions 148
References 152
7 Optical Flow in Real-Time Systems and Autonomous Vehicles 153
7.1 Introduction 153
7.2 Role of Optical Flow in Real-Time Systems 154
7.3 Real-Time Optical Flow in Autonomous Vehicles 155
7.4 Sensor Fusion and Real-Time Performance Optimization 156
7.5 Challenges and Future Directions 157
7.6 Conclusion 158
Examples with Solutions 159
References 161
8 Drones and Aerial Surveillance for Optical Flow Estimation in Adverse Outdoor Conditions 163
8.1 Introduction 163
8.2 Optical Flow for Navigation and Object Tracking 164
8.3 Challenges Regarding Adverse Outdoor Conditions 166
8.4 Improving Optical Flow Estimation for Adverse Conditions 167
8.5 Challenges and Future Directions 168
8.6 Conclusion 169
References 169
9 Conclusions and Future Work 171
9.1 Conclusions 171
9.2 Future Work 173
Index 177

Dr. Sana Rao is an Assistant Professor in the Department of Information and Communication Technology at Bahauddin Zakariya University, Multan, Pakistan. Her research areas are computer vision and motion estimation. She has significant experience collaborating with public and private sector research groups.
Dr. Rao Kashif is an Assistant Professor in the Department of Computer Science and Engineering at NUML, Multan, Pakistan. Prior, he is CEO and Managing Director at the Center to Advance Level Research and Development, LTD, Pakistan.