Night Vision Processing and Understanding.pdf

发布时间：2022-06-10 发布人：admin 分类：说明书资料大小：14.35M 资料格式：pdf 举报版权申诉

weixin_38290023-11178905-16359647397169625679.pdf-第1页.png

第1页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第2页.png

第2页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第3页.png

第3页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第4页.png

第4页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第5页.png

第5页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第6页.png

第6页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第7页.png

第7页 / 共274页

weixin_38290023-11178905-16359647397169625679.pdf-第8页.png

第8页 / 共274页

Foreword by Huilin Jiang

Foreword by Xiangqun Cui

Preface

Acknowledgements

Contents

1 Introduction

1.1 Research Topics of Multidimensional Night-Vision Information Understanding

1.1.1 Data Analysis and Feature Representation Learning

1.1.2 Dimension Reduction Classification

1.1.3 Information Mining

1.2 Challenges to Multidimensional Night-Vision Data Mining

1.3 Summary

References

2 High-SNR Hyperspectral Night-Vision Image Acquisition with Multiplexing

2.1 Multiplexing Measurement in Hyperspectral Imaging

2.2 Denoising Theory and HTS

2.2.1 Traditional Denoising Theory of HTS

2.2.2 Denoising Bound Analysis of HTS with S Matrix

2.2.3 Denoising Bound Analysis of HTS with H Matrix

2.3 Spatial Pixel-Multiplexing Coded Spectrometre

2.3.1 Typical HTS System

2.3.2 Spatial Pixel-Multiplexing Coded Spectrometre

2.4 Deconvolution-Resolved Computational Spectrometre

2.5 Summary

References

3 Multi-visual Tasks Based on Night-Vision Data Structure and Feature Analysis

3.1 Infrared Image Super-Resolution via Transformed Self-similarity

3.1.1 The Introduced Framework of Super-Resolution

3.1.2 Experimental Results

3.2 Hierarchical Superpixel Segmentation Model Based on Vision Data Structure Feature

3.2.1 Hierarchical Superpixel Segmentation Model Based on the Histogram Differential Distance

3.2.2 Experimental Results

3.3 Structure-Based Saliency in Infrared Images

3.3.1 The Framework of the Introduced Method

3.3.2 Experimental Results

3.4 Summary

References

4 Feature Classification Based on Manifold Dimension Reduction for Night-Vision Images

4.1 Methods of Data Reduction and Classification

4.1.1 New Adaptive Supervised Manifold Learning Algorithms

4.1.2 Kernel Maximum Likelihood-Scaled LLE for Night-Vision Images

4.2 A New Supervised Manifold Learning Algorithm for Night-Vision Images

4.2.1 Review of LDA and CMVM

4.2.2 Introduction of the Algorithm

4.2.3 Experiments

4.3 Adaptive and Parameterless LPP for Night-Vision Image Classification

4.3.1 Review of LPP

4.3.2 Adaptive and Parameterless LPP (APLPP)

4.3.3 Connections with LDA, LPP, CMVM and MMDA

4.3.4 Experiments

4.4 Kernel Maximum Likelihood-Scaled Locally Linear Embedding for Night-Vision Images

4.4.1 KML Similarity Metric

4.4.2 KML Outlier-Probability-Scaled LLE (KLLE)

4.4.3 Experiments

4.4.4 Discussion

4.5 Summary

References

5 Night-Vision Data Classification Based on Sparse Representation and Random Subspace

5.1 Classification Methods

5.1.1 Research on Classification via Semi-supervised Random Subspace Sparse Representation

5.1.2 Research on Classification via Semi-supervised Multi-manifold Structure Regularisation (MMSR)

5.2 Night-Vision Image Classification via SSM–RSSR

5.2.1 Motivation

5.2.2 SSM–RSSR

5.2.3 Experiment

5.3 Night-Vision Image Classification via P-RSSR

5.3.1 Probability Semi-supervised Random Subspace Sparse Representation (P-RSSR)

5.3.2 Experiment

5.4 Night-Vision Image Classification via MMSR

5.4.1 MR

5.4.2 Multi-manifold Structure Regularisation (MMSR)

5.4.3 Experiment

5.5 Summary

References

6 Learning-Based Night-Vision Image Recognition and Object Detection

6.1 Machine Learning in IM

6.1.1 Autoencoders

6.1.2 Feature Extraction and Classifier

6.2 Lossless-Constraint Denoising Autoencoder Based Night-Vision Image Recognition

6.2.1 Denoising and Sparse Autoencoders

6.2.2 LDAE

6.2.3 Experimental Comparison

6.3 Integrative Embedded Night-Vision Target Detection System with DPM

6.3.1 Algorithm and Implementation of Detection System

6.3.2 Experiments and Evaluation

6.4 Summary

References

7 Non-learning-Based Motion Cognitive Detection and Self-adaptable Tracking for Night-Vision Videos

7.1 Target Detection and Tracking Methods

7.1.1 Investigation of Infrared Small-Target Detection

7.1.2 Moving Object Detection Based on Non-learning

7.1.3 Researches on Target Tracking Technology

7.2 Infrared Small Object Detection Using Sparse Error and Structure Difference

7.2.1 Framework of Object Detection

7.2.2 Experimental Results

7.3 Adaptive Mean Shift Algorithm Based on LARK Feature for Infrared Image

7.3.1 Tracking Model Based on Global LARK Feature Matching and CAMSHIFT

7.3.2 Target Tracking Algorithm Based on Local LARK Feature Statistical Matching

7.3.3 Experiment and Analysis

7.4 An SMSM Model for Human Action Detection

7.4.1 Technical Details of the SMSM Model

7.4.2 Experiments Analysis

7.5 Summary

References

8 Colourization of Low-Light-Level Images Based on Rule Mining

8.1 Research on Colorization of Low-Light-Level Images

8.2 Carm

8.2.1 Summary of the Principle of the Algorithm

8.2.2 Mining of Multi-attribute Association Rules in Grayscale Images

8.2.3 Colorization of Grayscale Images Based on Rule Mapping

8.2.4 Analysis and Comparison of Experimental Results

8.3 Multi-sparse Dictionary Colorization Algorithm Based on Feature Classification and Detail Enhancement

8.3.1 Colorization Based on a Single Dictionary

8.3.2 Multi-sparse Dictionary Colorization Algorithm for Night-Vision Images, Based on Feature Classification and Detail Enhancement

8.3.3 Experiment and Analysis

8.4 Summary

References

Lianfa Bai · Jing Han · Jiang Yue Night Vision Processing and Understanding

Night Vision Processing and Understanding

Lianfa Bai Jing Han Jiang Yue Night Vision Processing and Understanding 123

Jing Han School of Electronic and Optical Engineering Nanjing University of Science and Technology Nanjing, Jiangsu, China Lianfa Bai School of Electronic and Optical Engineering Nanjing University of Science and Technology Nanjing, Jiangsu, China Jiang Yue National Key Laboratory of Transient Physics Nanjing University of Science and Technology Nanjing, Jiangsu, China ISBN 978-981-13-1668-5 https://doi.org/10.1007/978-981-13-1669-2 ISBN 978-981-13-1669-2 (eBook) Library of Congress Control Number: 2018965447 © Springer Nature Singapore Pte Ltd. 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, speciﬁcally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microﬁlms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, in this publication does not imply, even in the absence of a speciﬁc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. trademarks, service marks, etc. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Foreword by Huilin Jiang this research method that combines intelligent With the continuous development of information mining, cognitive computing and other disciplines, information understanding and night-vision imaging technology can effectively simulate human perception mechanisms and processes, and has wide application prospects in the ﬁeld of night-vision information perception. However, due to the cross-disciplinary studies, some new developments and achievements are scattered. For the time being, there is less systematic introduction of books in this area, especially in the ﬁeld of night-vision technology. In order to change this situation and better promote the development of night-vision information technology, this book focuses on new theories and tech- nologies currently being developed in the ﬁelds of multispectral imaging, dimen- sionality reduction, data mining, feature classiﬁcation learning, target recognition, object detection, colorization algorithm, etc., by exploring optimization models and new algorithms. It solves the applications of perception computing, mining learning and information understanding technologies in night-vision data, and strives to demonstrate the major breakthroughs brought by modern information technology to the night-vision ﬁeld. This book is a landmark and timely contribution in this direction as it offers, for the ﬁrst time, detailed descriptions and analysis of this frontier theory and method of night-vision information processing. Based on the differences in the imaging environment, target characteristics and imaging methods, it concentrates on mul- tispectral data, video data, etc., and researches a variety of information mining and perceptual understanding algorithms, which aims to analyse new processing methods for multiple types of scenes and targets. The selection of content fully reﬂects the main technical connotations and dynamics of the new ﬁeld of night vision. Eight chapters include spectral imaging and coding noise reduction, multi-vision tasks based on data structure and feature analysis, feature classiﬁcation based on manifold dimension reduction, data clas- siﬁcation based on sparse representation and random subspace, target detection and recognition based on learning, motion detection and tracking based on non-learning, colorization of night-vision images based on rule mining, etc., which v

vi Foreword by Huilin Jiang cover the comprehensive research areas of artiﬁcial intelligence in night vision. These provided algorithm models and hardware systems can be used as the refer- ence basis for the general design, algorithm design and hardware design personnel of the photoelectric system. In this monograph, Lianfa Bai, Jing Han and Jiang Yue have brought together their work in Night Vision Processing and Understanding over the past decade to result in a book that will become a standard for the area. Well done. Changchun, China Huilin Jiang

Foreword by Xiangqun Cui Night-vision technology is used to extend human activities beyond the limits of natural visual ability. For example, it is widely used in the ﬁelds for observation, monitoring and low-light detection. Night-vision research includes low-level light (LLL) vision, infrared thermal imaging, ultraviolet imaging and active near-infrared systems. Multi-source night-vision technology uses the complementarity of multi-sensor information to solve the problem of incomplete or inaccurate infor- mation of single imaging sensor. However, the extraction of useful information from multi-sensor presents new problems. Thus, is necessary to synthesise information provided by different sensors. The possible redundancy and contra- diction of multi-source information can thus be eliminated, allowing users to describe complete and consistent target information in complex scenes. it Along with the advancement of information mining, cognitive computing and other disciplines, it is known and believed that combining night-vision technology and intelligent information understanding can effectively simulate human percep- tion. Data structure analyses, feature representation learning, dimension reduction classiﬁcations and information mining theories have all been studied extensively. These models and algorithms have obvious advantages over conventional methods in terms of information understanding. However, there are few studies on feature mining of night-vision images. For complex night-vision information processing, the manifold learning, classiﬁcation and data mining methods still require investigation. For practical applications (e.g. security, defence and industry), this research is needed to solve problems of multispectral target detection or large-number image classiﬁcation and recognition. This book compiles an intelligent understanding of night-vision data under high dimensionality and complexity. Several night-vision data processing methods, based on feature learning, dimension reduction classiﬁ- cation and information mining, are explored and studied, providing various new technical approaches for information detection and understanding. This book aims to present a systematic and comprehensive introduction of the latest theories and technologies for various aspects of night-vision information technologies. Speciﬁcally, it covers multispectral imaging, dimension reduction, vii

viii Foreword by Xiangqun Cui data mining, feature classiﬁcation learning, object recognition, object detection, colorization algorithm, etc. Additionally, the application of the up-to-date opti- mization models and algorithms (including perception computing, mining learning and information understanding technologies) is explored to night-vision data and demonstrates major breakthroughs in the ﬁeld. The reader of this book will get both, a fairly comprehensive overview of the ﬁeld of night-vision processing and understanding, reached in the last two decades. I am very proud to have had the opportunity to follow this development for almost 20 years. Enjoy reading this book as I did. Nanjing, China Xiangqun Cui

分享到：

赞收藏

资料库

Night Vision Processing and Understanding.pdf

相关推荐

音视频

热门标签

最新资料