Land Cover Classification of Remotely Sensed Images: A Textural Approach

ISBN-13: 9783030665944 / Angielski / Twarda / 2021 / 176 str.

S. Jenicka

Spis treści

ABSTRACT

i

ACKNOWLEDGEMENTS

iii

DEDICATION

v

TABLE OF CONTENTS

vi

LIST OF FIGURES

xi

LIST OF TABLES

xiv

LIST OF SYMBOLS AND ABBREVIATIONS

xvi

1

INTRODUCTION TO REMOTE SENSING

1

 

1.1

Basics of Remote Sensing

1

 

1.2

Resolution Characteristics of remotely sensed imagery data

7

 

1.3

Reflectance Characteristics of Remotely Sensed Imagery

9

 

1.4

Remote sensing applications

12

 

1.5

Types of remotely sensed images

 

2

INTRODUCTION TO TEXTURE

14

 

2.1

Basics of texture

14

 

2.2

Texture analysis

 

3

LITERATURE SURVEY

19

 

3.1

Introduction

19

 

3.2

Survey Papers on Texture Models

19

 

3.3

Texture Models used for Characterization of Images

26

 

 

3.3.1

Structural Texture Models

27

 

 

3.3.2

Statistical Texture Models

27

 

 

3.3.3

Spectral Models

30

 

 

3.3.4

Model based Texture Models

30

 

 

3.3.5

Fuzzy based Models

31

 

 

3.3.6

Combined (texture and colour) approach Models

 

 

3.4

Classifiers applied in texture based study

42

 

3.5

Distance measures in texture based study

45

4

A FEW EXISTING BASIC AND MULTIVARIATE TEXTURE MODELS

49

 

4.1

Multivariate Local Binary Pattern

49

 

4.2

Multivariate Local Texture Pattern

50

 

4.3

Gray Level Co-occurrence Matrix

51

 

4.4

Texture Spectrum

54

 

4.5

Discrete Local Texture Pattern

 

 

4.6

Local Derivative Pattern

 

 

4.7

MATLAB codes of basic texture models

 

5

TEXTURE BASED SEGMENTATION USING BASIC TEXTURE MODELS

77

 

5.1

Texture based classification

77

 

5.2

Texture based segmentation

78

 

5.3

k-Nearest Neighbour (k-NN) Classifier

 

 

5.4

Experimental data

 

 

5.5

Matlab codes for texture based segmentation

 

 

 

5.5.1

GLCM and minimum distance classifier

 

 

 

5.5.2

LBP and minimum distance classifier

 

6

TEXTURE BASED SEGMENTATION USING LBP WITH SUPERVISED AND UNSUPERVISED CLASSIFIERS

 

 

6.1

Texture Segmentation using LBP with Supervised Classifiers

78

 

 

6.1.1

LBP with fuzzy k-NN

 

 

 

6.1.2

LBP with SVM

 

 

 

6.1.3

LBP with ANFIS

 

 

 

6.1.4

LBP with ELM

 

 

 

6.1.5

LBP with HMM

 

 

6.2

Texture Segmentation using LBP with Unsupervised Classifiers

 

 

 

6.2.1

LBP with SOM

 

 

 

6.2.2

LBP with FCM

 

7

TEXTURE BASED CLASSIFICATION OF REMOTELY SENSED IMAGES

 

 

7.1

Issues and challenges in texture based classification of remotely sensed images

 

 

7.2

The proposed texture model

 

 

7.3

Matlab code : Classification Procedure for texture based classification of remotely sensed images using the proposed texture model

 

 

7.4

The proposed approach using HMM

 

8

PERFORMANCE METRICS

135

REFERENCES

 

LIST OF PUBLICATIONS BY AUTHOR

 

AUTHOR’S BIOGRAPHY

 

O autorze

Dr. S. Jenicka completed her under graduation in Computer Science and Engineering at Thiagarajar College of Engineering, Madurai, Tamil Nadu in 1994. Later she finished her post-graduation in the same discipline in 2009 from Manonmaniam Sundaranar University, Tirunelveli, Tamil Nadu. She completed a doctorate in Computer Science and Engineering in 2014. Her research work was on ‘Texture based classification of remotely sensed images’. Her interests include Satellite image processing and texture segmentation. 

This book is the offspring of the expertise gained by Dr. Jenicka through the research work. She has got several online conference and journal publications with citation index. She has got nearly 13 years of teaching experience in reputed institutions.

Blurb

The book introduces two domains namely Remote Sensing and Digital Image Processing. It discusses remote sensing, texture, classifiers, and procedures for performing the texture-based segmentation and land cover classification.  

The first chapter discusses the important terminologies in remote sensing, basics of land cover classification, types of remotely sensed images and their characteristics. The second chapter introduces the texture and  a detailed literature survey citing papers related to texture analysis and image processing. The third chapter describes basic texture models for gray level images and multivariate texture models for color or remotely sensed images with relevant Matlab source codes. The fourth chapter focuses on texture-based classification and texture-based segmentation. The Matlab source codes for performing supervised texture based segmentation using basic texture models and minimum distance classifier are listed. The fifth chapter describes supervised and unsupervised classifiers. The experimental results obtained using a basic texture model (Uniform Local Binary Pattern) with the classifiers described earlier are discussed through the relevant Matlab source codes. The sixth chapter describes land cover classification procedure using multivariate (statistical and spectral) texture models and minimum distance classifier with Matlab source codes. A few performance metrics are also explained. The seventh chapter explains how texture based segmentation and land cover classification are performed using the hidden Markov model with relevant Matlab source codes. The eighth chapter gives an overview of  spatial data analysis and other existing land cover classification methods. The ninth chapter addresses the research issues and challenges associated with land cover classification using textural approaches.  

This book is useful for  undergraduates in Computer Science and Civil Engineering and postgraduates who plan to do research or project work in digital image processing. The book can serve as a guide to those who narrow down their research to processing remotely sensed images. It addresses a wide range of texture models and classifiers. The book not only guides but aids the reader in implementing the concepts through the Matlab source codes listed. In short, the book will be a valuable resource for growing academicians to gain expertise in their area of specialization and students who aim at gaining in-depth knowledge through practical implementations. The exercises given under texture based segmentation (excluding land cover classification exercises) can serve as lab exercises for the undergraduate students who learn texture based image processing.