Granular Video Computing: With Rough Sets, Deep Learning And In Iot

Description

This volume links the concept of granular computing using deep learning and the Internet of Things to object tracking for video analysis. It describes how uncertainties, involved in the task of video processing, could be handled in rough set theoretic granular computing frameworks. Issues such as object tracking from videos in constrained situations, occlusion/overlapping handling, measuring of the reliability of tracking methods, object recognition and linguistic interpretation in video scenes, and event prediction from videos, are the addressed in this volume. The book also looks at ways to reduce data dependency in the context of unsupervised (without manual interaction/ labeled data/ prior information) training.This book may be used both as a textbook and reference book for graduate students and researchers in computer science, electrical engineering, system science, data science, and information technology, and is recommended for both students and practitioners working in computer vision, machine learning, video analytics, image analytics, artificial intelligence, system design, rough set theory, granular computing, and soft computing. Contents Preface About the Authors List of Figures List of Tables List of Algorithms 1. Introduction: Video Processing, Granular Computing, Rough Sets, Deep Learning and IoT 1.1 Introduction 1.2 Video Computing 1.2.1 Video Tracking 1.2.1.1 Extraction and Modeling of Moving Objects 1.2.1.2 Feature Selection for Tracking 1.2.1.3 Tracking of Moving Objects 1.2.1.4 Partially Supervised Tracking 1.2.1.5 Unsupervised Tracking 1.2.2 Trustability Measure of Tracking 1.2.3 Object Recognition from Scene 1.2.4 Video Understanding 1.3 Granular Computing 1.3.1 Granulation 1.3.2 Information Granules 1.4 Rough Sets 1.4.1 Definitions 1.4.2 Reduct and Core 1.4.3 Partial Dependency in Knowledge Base 1.4.4 Indiscernibility Matrix 1.4.5 Neighborhood Rough Sets 1.4.6 Rough Sets and Image Processing 1.4.7 Rough Sets and Uncertainty Modeling 1.4.7.1 Image as Rough Set 1.4.8 Rough Entropy 1.5 Deep Learning 1.5.1 Deep Learning and Image Processing 1.5.2 Deep Learning Models 1.6 Internet of Things (IoT) 1.6.1 IoT and Video Processing 1.6.2 IoT: Features and Benefits 1.6.3 IoT Applications 1.7 Scope and Organization of the Book Bibliography 2. Partial Supervised Tracking 2.1 Introduction 2.2 Popular Approaches for Partially Supervised Tracking 2.3 Rough Entropy-based Spatial Segmentation 2.3.1 Formation of Granules 2.3.2 Algorithm for Optimal Thresholding for Spatial Segmentation 2.4 Background Estimation-based Temporal Segmentation 2.4.1 Three-point Estimation Technique 2.5 Target Localization and Tracking: Algorithm 2.6 Experimental Results 2.6.1 Results of Spatial Segmentation 2.6.2 Results of Temporal Segmentation 2.6.3 Results of Reconstruction 2.6.4 Results of Tracking 2.7 Conclusions and Discussions Bibliography 3. Unsupervised Tracking 3.1 Introduction 3.2 Popular Methods for Unsupervised Tracking 3.3 Neighborhood Granular Rough Rule-base and Flow Graph for Unsupervised Tracking (NRBFG) 3.3.1 Formation of Neighborhood Granules 3.3.1.1 Spatio-color Neighborhood Granules 3.3.1.2 3-D Spatio-temporal Neighborhood Granules 3.3.1.3 Color Neighborhood Granules 3.4 Rough Rule-base and Unsupervised Tracking 3.4.1 Acquiring Features for Initial Unsupervised Labeling 3.4.2 Rule-base in Pixel Level 3.4.3 Rule-base in Granular Level 3.5 Adaptive Rule-base: A Granular Flow Graph Approach 3.5.1 Information Flow Graph 3.5.2 Flow Graph in Video: Relevance and Issues 3.5.3 Expected Flow Graph 3.5.4 Mutual Dependency Between Flow Graphs 3.6 Experimental Results 3.6.1 Results of Tracking 3.6.1.1 Comparisons Between Neighborhood Granular Level and Pixel Level Methods 3.6.1.2 Comparisons Between with Flow Graph and Without Flow Graph 3.6.2 Comparisons 3.7 Conclusions and Discussions Bibliography 4. Unsupervised Occlusion Handling 4.1 Introduction 4.2 Related Work on Handling Overlapping and Occlusion 4.3 Unsupervised Methodology for Handling Overlapping and Occlusion 4.3.1 Overview 4.3.2 Formation of Spatio-color Neighborhood Granules 4.3.3 NRS Filter for Unsupervised Object Estimation 4.3.4 Performance Measure of NRS Filter 4.3.5 Temporal Neighborhood Granules and Location Estimation 4.3.5.1 Velocity Neighborhood Granules (ℵV ) 4.3.5.2 Acceleration Neighborhood Granule (ℵA) 4.3.6 Roughness in Object Approximation 4.3.7 Intuitionistic Entropy 4.3.7.1 Neighborhood Rough Entropy 4.3.7.2 Neighborhood Probabilistic Entropy 4.4 Algorithm for Tracking 4.5 Experimental Results 4.5.1 Selection of Parameters and Initial Assumptions 4.5.2 Effectiveness of Unsupervised Labeling and Estimation 4.5.3 Results of Tracking 4.6 Conclusions and Discussions Bibliography 5. Trustability Measures of Tracking Algorithms 5.1 Introduction 5.2 Different Performance Evaluation Metrics 5.3 Indices Without Ground Truth Information 5.3.1 RGB-based Evaluation Indices 5.3.1.1 k-index 5.3.1.2 f-index 5.3.1.3 b-index 5.3.2 Tracking Indices With Depth Feature 5.3.2.1 ED Index: Foreground Segmentation 5.3.2.2 ℵI Index: Mis-classification 5.4 Characteristic Differences Among the Unsupervised Measures 5.4.1 Effectiveness of RGB-based Measures: Evaluation of Mis-tracked Frames 5.4.2 Validation of Depth-based Measures 5.5 Trustability Measure of Different Methods Under Different Scenarios 5.5.1 Trustability for Single Object Tracking 5.5.2 Trustability for Multiple Object Tracking 5.5.3 Trustability for Tracking with Occlusion/Overlapping 5.6 Conclusions and Discussions Bibliography 6. Object Recognition and Deep Learning 6.1 Introduction 6.2 Granulated Deep Learning 6.2.1 Granulation Techniques 6.2.1.1 Image Definition with Rough Sets 6.2.1.2 Formation of Granules 6.2.2 Conventional Convolution Neural Network 6.2.3 Granulated Deep Learning: Concepts, Characteristics and Relevance 6.2.4 Granulated Deep Learning: Object Tracking and Recognition 6.3 Experimental Results 6.3.1 Results of Moving Object Detection and Tracking 6.3.2 Comparative Study 6.4 Z-number-based Scene Description 6.4.1 Formulation of Measures 6.4.2 Significance of Z-number-based Measure 6.5 Conclusions and Discussions Bibliography 7. Video Conceptualization 7.1 Introduction 7.2 Related Work 7.3 Video Conceptualization and IoT Architecture 7.3.1 Formation of Two-layered Motion Granules 7.3.1.1 First Layer: Spatio-temporal Granules 7.3.1.2 Second Layer: Motion Granules 7.3.2 Object–Background as Rough Sets 7.3.3 Motion Entropy 7.4 Conceptualizing the Video by Estimating the Moving Patterns of Objects and Precognized Events 7.5 Results and Discussions 7.5.1 Preliminary Assumptions and Parameter 7.5.2 Experimental Results in PC 7.5.3 Comparative Results 7.5.4 Experimental Results in IoT 7.6 Conclusions and Discussions Bibliography Index