Computer Vision and Image Processing: 7th International Conference, CVIP 2022, Nagpur, India, November 4–6, 2022, Revised Selected Papers, Part II

Description

This two volume set (CCIS 1776-1777) constitutes the refereed proceedings of the 7th International Conference on Computer Vision and Image Processing, CVIP 2022, held in Nagpur, India, November 4–6, 2022. The 110 full papers and 11 short papers were carefully reviewed and selected from 307 submissions. Out of 121 papers, 109 papers are included in this book. The topical scope of the two-volume set focuses on Medical Image  Analysis,  Image/  Video  Processing  for  Autonomous  Vehicles,  Activity Detection/  Recognition,  Human  Computer  Interaction,  Segmentation  and  Shape Representation,  Motion  and  Tracking,  Image/  Video  Scene  Understanding,  Image/Video  Retrieval,  Remote  Sensing,  Hyperspectral  Image  Processing,  Face,  Iris, Emotion, Sign Language and Gesture Recognition, etc. Preface Organization Contents – Part II Contents – Part I An Efficient Residual Convolutional Neural Network with Attention Mechanism for Smoke Detection in Outdoor Environment 1 Introduction 2 Related Work 3 Proposed Model 4 Results and Discussion 4.1 Dataset 4.2 Evaluation Measures 4.3 Ablation Study 4.4 Performance Evaluation 5 Conclusion References Expeditious Object Pose Estimation for Autonomous Robotic Grasping 1 Introduction 2 Overview of the Pose Estimation Pipeline 2.1 Phases of the Approach 2.2 Synthetic Data Collection 3 Pose Estimation Models 3.1 Model-1: UnityVGG16 3.2 Model-2: Pose6DSSD 3.3 Model-3: DOSSE-6D 3.4 Model-4: AHR-DOSSE-6D 4 Training Configuration and Model Evaluation 4.1 Experimental Setup 4.2 Loss Functions 4.3 Optimizer and Evaluation Details 5 Results 5.1 Unity Simulation Scenarios 5.2 LINEMOD Dataset 6 Conclusion References SRTGAN: Triplet Loss Based Generative Adversarial Network for Real-World Super-Resolution 1 Introduction 2 Related Works 3 Proposed Method 3.1 Loss Functions 4 Experimental Results 4.1 Training Details 4.2 Ablation Study 4.3 Quantitative Analysis 4.4 Qualitative Analysis 5 Limitations 6 Conclusion References Machine Learning Based Webcasting Analytics for Indian Elections - Reflections on Deployment 1 Introduction 2 Related Work 3 Proposed Methodology 3.1 Real Time Streaming Protocol (RTSP) 3.2 You Look Only Once (YOLO) 3.3 DeepSORT Tracking Framework 3.4 Multithreading for Enhancing Execution Speed 3.5 REST APIs for Analytics Data Transfer 4 Deployment and Results 4.1 System Architecture 4.2 Practical Challenges 5 Conclusion and Future Work References Real-Time Plant Species Recognition Using Non-averaged DenseNet-169 Deep Learning Paradigm 1 Introduction 2 Related Works 2.1 Conventional Methods 2.2 Deep Learning Methods 3 Methodology 3.1 Datasets 3.2 Preprocessing of Dataset Images 3.3 NADenseNet-169 CNN Model 3.4 Training Algorithm 3.5 Hardware Setup and Software Tools 4 Results and Discussion 4.1 Performance Measure of NADenseNet-169 Model 4.2 NADenseNet-169 Architecture with GAP Layer 4.3 Real-Time Prediction of Plant Species 5 Conclusion References Self Similarity Matrix Based CNN Filter Pruning 1 Introduction 2 Related Works 3 Proposed Method 3.1 Self-similarity Matrix 3.2 Greedy Method 3.3 Area Based Method 4 Observation and Results 4.1 ResNet Pruning 4.2 VGG Pruning 5 Conclusion References Class Agnostic, On-Device and Privacy Preserving Repetition Counting of Actions from Videos Using Similarity Bottleneck 1 Introduction 2 Related Work 3 Method 3.1 Data Augmentation 3.2 Architecture 4 Experiments 4.1 Architecture Variations 4.2 Training and Evaluation 4.3 Results 5 Conclusion References Vehicle ReID: Learning Robust Feature Using Vision Transformer and Gradient Accumulation for Vehicle Re-identification 1 Introduction 2 Related Work 3 Methodology 3.1 Overview 3.2 Vision Transformer 3.3 Gradient Accumulation 3.4 Side Information Embedding 3.5 Optimization Objective 4 Experiments 4.1 Dataset 4.2 Evaluation Metric 4.3 Implementation Details 4.4 Results and Discussion 5 Conclusion References Attention Residual Capsule Network for Dermoscopy Image Classification 1 Introduction 2 Methodology 2.1 Attention Residual Capsule Network (ARCN) 2.2 Network Training with Imbalanced Data 3 Experimental Protocol 3.1 Dataset 3.2 Evaluation Metrics 3.3 Implementation Details 4 Result and Discussion 5 Conclusion References SAMNet: Semantic Aware Multimodal Network for Emoji Drawing Classification 1 Introduction 2 Related Works 3 Proposed Method 3.1 Data 3.2 SAMNet Architecture 4 Experiments and Results 4.1 Architecture Evaluation 4.2 DigitalInk Evaluation 5 Conclusion References Segmentation of Smoke Plumes Using Fast Local Laplacian Filtering 1 Introduction and Background 2 Smoke Plume Data Collection 3 Proposed Methodology 3.1 Segmentation Using the Fast Local Laplacian Filter (FLLF) Technique 3.2 Deep Learning-Based Smoke/Plume Segmentation with FLLF Training Samples 4 Analysis of Results 4.1 Quantitative Metrics of Evaluation 4.2 Smoke Segmentation Results Using the FLLF Technique 4.3 Results Using UNet-Based Image Segmentation with the FLLF Technique 5 Conclusions 6 Future Work References Rain Streak Removal via Spatio-Channel Based Spectral Graph CNN for Image Deraining 1 Introduction 2 Proposed Methodology 2.1 Loss Function 3 Experiments 3.1 Implementation Details 3.2 Baseline Methods 3.3 Datasets 4 Results and Discussions 5 Conclusion References Integration of GAN and Adaptive Exposure Correction for Shadow Removal 1 Introduction 2 Proposed Method 2.1 Generator and Discriminator Learning 2.2 Adaptive Exposure Correction Module 2.3 Objectives and Loss Functions 2.4 Network Architecture and Training Strategy 2.5 Benchmark Dataset Adjustment 3 Experimental Results 4 Conclusion References Non-invasive Haemoglobin Estimation Using Different Colour and Texture Features of Palm 1 Introduction 2 Proposed Work 2.1 Data Acquisition 2.2 Dominant Frame Extraction from Palm Video 2.3 Colour Space Representations 2.4 Feature Extraction 2.5 Feature Selection 2.6 Prediction Model 3 Evaluation 4 Conclusion and Future Scope References Detection of Coal Quarry and Coal Dump Regions Using the Presence of Mine Water Bodies from Landsat 8 OLI/TIRS Images 1 Introduction 1.1 Objectives 2 Background Techniques 2.1 Coal Mine Index (CMI) 2.2 Modified Normalized Difference Water Index 2.3 Bare Soil Index (BI) 2.4 Morphological Opening 2.5 Single Class SVM 3 Methodology 3.1 Detection of Quarry Regions with Water Bodies 3.2 Detection of Coal Quarry and Dump Using Single Class SVM 4 Data and Study Area 5 Results 5.1 Spectral Validation 5.2 Performance Analysis 6 Conclusion and Future Aspect References Brain Tumor Grade Detection Using Transfer Learning and Residual Multi-head Attention Network 1 Introduction 2 Methodology 2.1 Transfer Learning 2.2 Residual Multi-head Attention Network 2.3 Proposed Model 3 Results and Discussion 3.1 Dataset Description 3.2 Experimental Setup 3.3 Performance Analysis 3.4 Comparison and Discussion 4 Conclusions References A Curated Dataset for Spinach Species Identification 1 Introduction 2 Materials and Methods 2.1 Spinach Leaf Database – Subset of MepcoTropicLeaf 2.2 Custom Made Deep Architecture 2.3 Transfer Learning Approach 3 Experiments and Discussions 4 Conclusion References Mobile Captured Glass Board Image Enhancement 1 Introduction 2 Related Work 3 The Proposed Algorithm 3.1 Highlight Removal 3.2 Image Segmentation 3.3 Selection of Proper Window 3.4 Color Assignment and Enhancement 4 Experiments 4.1 Datasets 4.2 Evaluation Measure 4.3 Quantitative Results Analysis 4.4 Qualitative Results 5 Conclusions References Computing Digital Signature by Transforming 2D Image to 3D: A Geometric Perspective 1 Introduction 1.1 Why 3D Reconstruction and Digital Signature? 1.2 Motivation and Applications 1.3 Related Work 1.4 Organization of the Paper 2 Proposed Method 3 Results and Experiments 4 Conclusion and Future Work References Varietal Classification of Wheat Seeds Using Hyperspectral Imaging Technique and Machine Learning Models 1 Introduction 2 Materials and Methods 2.1 Samples Selection and Preparation 2.2 Hyperspectral Imaging System 2.3 Hyperspectral Image Acquisition and Correction 3 Proposed Methodology 3.1 Machine Learning Models 4 Experiment and Results 4.1 Software Tools 4.2 Performance Evaluation Matrix 4.3 Analysis of Spectral Bands 4.4 Development of Classification Model 4.5 Comparison with State-of-the-art Methods 5 Conclusion and Future Work References Customized Preview Video Generation Using Visual Saliency: A Case Study with Vision Dominant Videos 1 Introduction 2 Proposed Method using Visual Saliency for Preview Video Generation 2.1 Phase Spectrum 2.2 Quaternion Representation of an Image 2.3 Opponent Color Scheme 2.4 Saliency Features Using Opponent Color Scheme 2.5 Saliency Map 3 Saliency Curve 4 Key Frame Extraction 5 Preview Generation 6 Block Diagram 7 Experimental Results and Analysis 7.1 Qualitative Analysis 7.2 Quantitative Analysis 7.3 Subjective Analysis 8 Related Work and Comparative Analysis 9 Limitations 10 Conclusion and Future Scope References Deep Dilated Convolutional Network for Single Image Dehazing 1 Introduction 2 Related Work 3 Method 3.1 Architecture 3.2 Dilated Convolution 3.3 Computational Complexity 3.4 Loss Function 3.5 Visualization of the Network 4 Experiments 4.1 Dataset 4.2 Training Details 4.3 Metrics and Compared Methods 4.4 Quantitative Metrics 4.5 Qualitative Analysis 4.6 Run Time 5 Conclusion References T2CI-GAN: Text to Compressed Image Generation Using Generative Adversarial Network 1 Introduction 2 Preliminaries 2.1 JPEG Compression 2.2 Generative Adversarial Network (GAN) 2.3 GloVe Model 3 Proposed Methodology 3.1 Network Architecture 3.2 Proposed T2CI-GAN Model-1: Training with JPEG Compressed DCT Images 3.3 Training T2CI-GAN Model-1 3.4 Proposed T2CI-GAN Model-2: With Modified Generator and Training with RGB Images 4 Experimental Results 4.1 Oxford-102 Flowers Dataset 4.2 Text to Compressed Image Results 4.3 Quantitative Evaluation of the Model 4.4 Comparative Study 5 Conclusion References Prediction of Fire Signatures Based on Fractional Order Optical Flow and Convolution Neural Network 1 Introduction 2 Methodology 2.1 Proposed Fractional Order Variational Optical Flow Model 2.2 Minimization 2.3 Binary Mask 2.4 Feature Extraction: Deep CNN Architecture 3 Experiments, Results and Discussion 3.1 Datasets 3.2 Performance Metrics 3.3 Experimantal Discussion 4 Conclusion References Colonoscopy Polyp Classification Adding Generated Narrow Band Imaging 1 Introduction 2 Literature Review 2.1 Polyp Classification 2.2 Image-to-Image Translation 3 Proposed Method 3.1 Proposed WL to NBI Translation Using CycleGAN 3.2 Polyp Classification Using Generated NBI Images 4 Result and Experimental Details 4.1 Datasets 4.2 Data Pre-processing 4.3 Implementation and Training Details 5 Conclusion References Multi-class Weather Classification Using Single Image via Feature Fusion and Selection 1 Introduction 2 Related Work 3 Proposed Method 3.1 Feature Extraction 3.2 Feature Selection 3.3 Classification 3.4 Implementation &Data Set Details 4 Experiments Results 4.1 Evaluation of Classifier 4.2 Classification Using Various Features 4.3 Accuracy & Feature Vector Size 4.4 Comparison with Other Methods 4.5 Evaluation Metrics 5 Conclusion References Scene Text Detection with Gradient Auto Encoders 1 Introduction 2 Related Work 3 Dataset Description 4 Proposed Method 4.1 Phase 1: Training 4.2 Phase 2: Testing 5 Experimental Results 6 Conclusion References A Novel Scheme for Adversarial Training to Improve the Robustness of DNN Against White Box Attacks 1 Introduction 2 Related Work 3 Proposed Defense Method 3.1 Cross-entropy Loss Function 3.2 Proposed Loss Function 3.3 Model Training 4 Adversarial Attack 4.1 Fast Gradient Sign Method (FGSM) 4.2 Projected Gradient Descent (PGD) 4.3 Carlini and Wagner (CW) Attack 5 Results 5.1 Experimental Settings 5.2 Results and Analysis 5.3 Comparison with Existing Defenses 5.4 Identifying Obfuscated Gradients 6 Conclusion References Solving Diagrammatic Reasoning Problems Using Deep Learning 1 Introduction 1.1 Related Work 1.2 Contributions 2 Proposed Framework 2.1 Encoder-Decoder Training 2.2 LSTM Training 3 Experiments and Results 3.1 Dataset 3.2 Accuracy 3.3 Model Predictions 4 Conclusion References Bird Species Classification from Images Using Deep Learning 1 Introduction 2 Literature Survey 3 Methodology 3.1 Dataset Description 3.2 Model Architecture 3.3 Other Models Used 4 Results and Analysis 4.1 Training 4.2 Model Loss Trend 4.3 Result 4.4 Comparison with State-of-Art 5 Conclusion and Future Work References Statistical Analysis of Hair Detection and Removal Techniques Using Dermoscopic Images 1 Introduction 2 Related Work 3 Hair Removal Techniques 4 Implementation Details 5 Result Analysis and Discussion 6 Conclusion References Traffic Sign Detection and Recognition Using Dense Connections in YOLOv4 1 Introduction 2 Related Work 3 Proposed Work 3.1 YOLOv4 Model 3.2 Dense Connection in Detection Neck 4 Experiment Results and Analysis 4.1 Experimental Setup 4.2 Evaluation Metrics 4.3 Dataset 4.4 Results and Analysis 5 Conclusion and Future Work References Novel Image and Its Compressed Image Based on VVC Standard, Pair Data Set for Deep Learning Image and Video Compression Applications 1 Introduction 2 Database Capturing and Characteristic 3 Results and Discussions 4 Conclusion References FAV-Net: A Simple Single-Shot Self-attention Based ForeArm-Vein Biometric 1 Introduction 2 Related Works 2.1 Vein-Based Traditional Biometrics 2.2 Vein-Based CNN Biometrics 2.3 Few-Shot Biometrics 2.4 Self-attention Biometrics 3 Proposed Approach 3.1 Base Network 3.2 Data Augmentation 3.3 Self-attention Mechanism 3.4 Loss Function and Training Strategy 3.5 Implementation Details 4 Experiments and Results 4.1 Database 4.2 Results 4.3 Ablation Study 4.4 Loss Functions 4.5 Other Applications 5 Conclusion and Future Work References Low-Textural Image Registration: Comparative Analysis of Feature Descriptors 1 Introduction 2 Motivation 3 Related and Relevant Work 4 Hardware Configuration and Experimental Setup 5 DataSet Preparation 6 Understanding Feature Descriptors 6.1 Scale Invariant Feature Transform (SIFT) 6.2 Speeded Up Robust Feature (SURF) 6.3 Oriented Fast and Rotated BRIEF (ORB) 6.4 Binary Robust Invariant Scalable Keypoints (BRISK) 6.5 Accelerated-KAZE (AKAZE) 7 Performance Comparison of Feature Descriptor Algorithms 8 Comparison of Registration Based on the Various Feature Descriptors 9 Conclusions References Structure-Based Learning for Robust Defense Against Adversarial Attacks in Autonomous Driving Agents 1 Introduction 2 Related Work 3 Research Design and Proposed Methodology 3.1 Autonomous Vehicular System Prototype 3.2 Deep Learning Based Classifier for Vehicle Navigation 3.3 Adversarial Attacks: Configuration, and Effects 3.4 Challenges in Real Time Data Acquisition 3.5 The Defense Strategy 4 Experimental Results and Discussion 4.1 Data Acquisition and Splitting 4.2 Performance Evaluation 5 Conclusion References Improvising the CNN Feature Maps Through Integration of Channel Attention for Handwritten Text Recognition 1 Introduction 2 Related Works 3 Channel Attention in HTR Architectures 3.1 Squeeze and Excite 3.2 Efficient Channel Attention 3.3 Combining Global CCI and Local CCI 4 Results and Discussion 4.1 Dataset 4.2 Experimental Setup 4.3 Evaluation Metrics 4.4 Results 5 Conclusion References XAIForCOVID-19: A Comparative Analysis of Various Explainable AI Techniques for COVID-19 Diagnosis Using Chest X-Ray Images 1 Introduction 2 Related Work 3 Explainable AI 3.1 LIME 3.2 Occlusion 3.3 Saliency 3.4 LRP 3.5 Deconvolution 3.6 GradCAM++ 3.7 AlbationCAM 3.8 XGrad-CAM 4 Experiments Setup 4.1 Data-set 4.2 Data Pre-processing 4.3 Network Model 4.4 Model Training 4.5 Interpretability Pipeline 5 Results and Interpretability Analysis 5.1 Qualitative Analysis 5.2 Quantitative Analysis 6 Discussions and Conclusion References Features Assimilation via Three-Stream Deep Networks for Spam Attack Detection from Images 1 Introduction 2 Literature Review 3 Proposed Method 3.1 Pre-processing 4 Experiments 4.1 Datasets 4.2 Multidataset Unification 4.3 Ablation Study 4.4 Quantitative Analysis Comparison with Other Methods 5 Conclusion References A Transformer-Based U-Net Architecture for Fast and Efficient Image Demoireing 1 Introduction 2 Related Work 3 Proposed Method 3.1 TrANSConv Block 3.2 Feed Forward Net 3.3 Restoration Modulator 4 Experimental Results 4.1 Dataset and Training Details 4.2 Result Analysis 5 Conclusion References Auxiliary CNN for Graspability Modeling with 3D Point Clouds and Images for Robotic Grasping 1 Introduction 2 Related Work 3 Method 3.1 GraspNet 3.2 Our Method 4 Experimental Results 4.1 Dataset 4.2 Evaluation 4.3 Quantitative Analysis 4.4 Qualitative Analysis 4.5 Ablation Study 5 Conclusion References Face Presentation Attack Detection Using Remote Photoplethysmography Transformer Model 1 Introduction 2 Proposed Face PAD Method 2.1 Cropping Face Regions 2.2 PhysFormer 2.3 Classification: 3DCNN Network 3 Experiments and Results 3.1 Dataset 3.2 Visualisation of rPPG Signals Extracted from Different Face Regions 3.3 Quantitative Results of the Proposed PAD Model 4 Conclusions References MTFL: Multi-task Federated Learning for Classification of Healthcare X-Ray Images 1 Introduction 2 Literature Survey 3 Methodology 3.1 Multi-task Learning 3.2 Federated Learning 3.3 Multi-task Federated Learning 4 Experimental Setup 4.1 Dataset 4.2 Data Augmentation 4.3 Global/Local Model 4.4 Construction of the Clients 4.5 Model Setting 4.6 Implementation Framework 4.7 Performance Measures 5 Results and Discussion 5.1 Results on Chestx-Ray Dataset 5.2 Results on Tuberculosis (TB) Dataset 5.3 Discussion 6 Conclusion References Speech-Based Automatic Prediction of Interview Traits 1 Introduction 2 Literature Review 3 Proposed Methodology 3.1 Feature Extraction 3.2 Methodology 4 Experiments and Results 4.1 Dataset 4.2 Experimental Settings 4.3 Performance Metrics 4.4 Models 4.5 Ablation Study 5 Results and Discussion 5.1 Comparison with the State-of-the-Art (SOTA) Approaches 5.2 Discussion 6 Conclusion and Future Work References Pneumonia Detection Using Deep Learning Based Feature Extraction and Machine Learning 1 Introduction 2 Related Works 3 Methodology 3.1 The Proposed Architecture 4 Experimental Setup, Results and Comparative Analysis 4.1 Dataset 4.2 Evaluation Metrics 4.3 Results 5 Conclusion References Classification of Synthetic Aperture Radar Images Using a Modified DenseNet Model 1 Introduction 2 Background 2.1 Existing Deep Learning Based Works on SAR Image Classification 2.2 DenseNet Model and Its Variants 3 Experimental Evaluation of the Different DenseNet Variants 3.1 Implementation 3.2 Results and Observations 4 Proposed Work 4.1 Implementation, Results and Discussions 5 Conclusion References Combining Non-local Sparse and Residual Channel Attentions for Single Image Super-resolution Across Modalities 1 Introduction 1.1 Contribution 2 Related Works 3 Proposed Architecture 3.1 Non Local Sparse Attention Module 3.2 Residual Channel Attention Module 3.3 Loss Function 4 Experimental Results 4.1 Training and Implementation Details 4.2 Results and Comparison 5 Results on Other Image Modalities 5.1 Depth Map 5.2 Medical Images 6 Conclusion References An End-to-End Fast No-Reference Video Quality Predictor with Spatiotemporal Feature Fusion 1 Introduction 2 Proposed VQA Model 3 Experiments 3.1 VQA Databases 3.2 Experimental Settings and Performance Criteria 3.3 Performance Analysis 4 Conclusion References Low-Intensity Human Activity Recognition Framework Using Audio Data in an Outdoor Environment 1 Introduction 2 Related Literature 3 Dataset 4 Methodology 4.1 Problem Formulation 4.2 Data Preprocessing 4.3 Deep Model 5 Experiments and Result Analysis 5.1 Implementation Details 5.2 Comparison with Existing Works 5.3 Analysis of Confusion Matrix 6 Conclusion References Detection of Narrow River Trails with the Presence of Highways from Landsat 8 OLI Images 1 Introduction 1.1 Related Works 1.2 Objectives and Contributions 2 Methodology 2.1 Spectral Index 2.2 Enhancing Curvilinear Patterns 2.3 Removal of Straight Lines 2.4 Morphological Operations 2.5 Connected Component Analysis 3 Data and Study Area 4 Results and Discussion 4.1 Validation 5 Conclusions References Unsupervised Image to Image Translation for Multiple Retinal Pathology Synthesis in Optical Coherence Tomography Scans 1 Introduction 2 Related Work 3 Methodology 3.1 Framework 3.2 Losses 4 Experiments and Results 4.1 Dataset Description 4.2 Experimental Setup and Results 5 Conclusion and Future Work References Combining Deep-Learned and Hand-Crafted Features for Segmentation, Classification and Counting of Colon Nuclei in H&E Stained Histology Images*-12pt 1 Introduction 2 Methodology 2.1 Data Pre-processing and Augmentation 2.2 HoVer-Net Baseline Model 2.3 Proposed Model 2.4 Handcrafted Feature Descriptors 3 Experimental Details 3.1 Implementation and Parameter Settings 3.2 Quantitative Performance Metrics 4 Results and Discussion 5 Conclusion References Multiple Object Tracking Based on Temporal Local Slice Representation of Sub-regions 1 Introduction 2 Methods and Materials 2.1 Problem Statements 2.2 Dataset 2.3 Verification of Prediction 2.4 Experimental Setup 2.5 Details of Technical Implementation 2.6 Experimental Set up 2.7 Flow Diagrams 2.8 Details of the Final Experiment 3 Results and Discussion 4 Conclusions and Future Directions References An Ensemble Approach for Moving Vehicle Detection and Tracking by Using Ni Vision Module 1 Introduction 2 A Designed Framework for Ensemble Approach for Moving Vehicle Detection and Tracking 2.1 Automatic Vehicle Detection Using Vision Assistant and Tracker Module 2.2 Automatic Vehicle Detection Using Labview Based Vision Assistant 3 Result and Discussion 4 Conclusion and Future Scope References Leaf Spot Disease Severity Measurement in Terminalia Arjuna Using Optimized Superpixels*-12pt 1 Introduction 2 Related Works 3 Materials and Methods 3.1 Simple Linear Iterative Clustering (SLIC) 3.2 Multi-objective Cuckoo Search 3.3 Objective Functions 3.4 Color Moment Features 3.5 Dataset 4 Proposed Method 5 Results and Analysis 6 Conclusion References Author Index