Multispectral Image Analysis Using the Object-Oriented Paradigm

Preface

This book is intended for students, research scientists, and professionals in the remote sensing industry who have a basic understanding of remote sensing principles, image processing, and applications of remotely sensed data. This book will appeal to users who apply imagery to a variety of mapping applications, including vegetation mapping, identification of man-made structures from imagery, mapping of urban growth patterns, and other applications. I wrote this book keeping in mind users with diverse educational backgrounds. 

This book is designed to explore the new object-oriented paradigm of image analysis. The content is tailored to both novice users as well as to advanced users who are ready to use or have already tried using objects for image analysis. I cover a variety of applications and demonstrate the advantages of object-based image analysis, using step-by-step examples and, wherever necessary, explaining the functionality in eCognition to accomplish tasks.

Object-based image analysis is a paradigm shift as compared to traditional pixel-based image analysis approaches and brings a fresh, new perspective to the remote sensing discipline. Advantages of this approach are demonstrated using various hands-on exercises in this book. For example, it is efficient to identify an agriculture crop based on vegetation boundaries rather than every pixel within the field. 

The spatial and spectral variations within a field, due to vegetation density, crop stress, presence of irrigation, and ditches that run through the field, make classification of a crop challenging on a pixel-by-pixel basis. The new object-oriented paradigm creates objects by grouping pixels with similar spectral and spatial characteristics and reduces crop classification problems to a few objects rather than thousands of pixels. This approach lets you take advantage of the true power of remote sensing

by exploiting all the dimensions of remote sensing, including spectral, spatial, contextual, morphological, and temporal aspects for information extraction.

Table of Contents


Preface
List of Figures
List of Tables
Chapter 1 Introduction ..........................................................................................1
1.1 Background ......................................................................................................1
1.2 Objects and Human Interpretation Process .....................................................1
1.2.1 Human Interpretation Process versus Computer-Aided
Image Analysis.....................................................................................2
1.3 Object-Oriented Paradigm ...............................................................................3
1.4 Organization of the Book ................................................................................3
Chapter 2 Multispectral Remote Sensing.............................................................5
2.1 Spatial Resolution ............................................................................................5
2.2 Spectral Resolution ..........................................................................................9
2.3 Radiometric Resolution..................................................................................10
2.4 Temporal Resolution ......................................................................................11
2.5 Multispectral Image Analysis ........................................................................12


Chapter 3 Why an Object-Oriented Approach? .................................................15
3.1 Object Properties............................................................................................16
3.2 Advantages of an Object-Oriented Approach ...............................................18
Chapter 4 Creating Objects.................................................................................19
4.1 Image Segmentation Techniques ...................................................................19
4.1.1 Public Domain Image Segmentation Software .................................20
4.1.2 eCognition Segmentation...................................................................20
4.2 Creating and Classifying Objects at Multiple Scales....................................28
4.3 Object Classification ......................................................................................30
4.4 Creating Multiple Levels ...............................................................................36
4.5 Creating Class Hierarchy and Classifying Objects .......................................39
4.6 Final Classification Using Object Relationships between Levels.................43
43064_C000.fm Page v Wednesday, November 1, 2006 8:52 AM

Chapter 5 Object-Based Image Analysis............................................................47
5.1 Image Analysis Techniques............................................................................47
5.1.1 Unsupervised Classification...............................................................47
5.1.2 Supervised Classification ...................................................................48
5.1.3 Rule-Based Classification ..................................................................49
5.1.4 Classification and Regression Trees (CART) and Decision Trees ...56
5.1.5 Neural Nets and Fuzzy Logic Classification.....................................61
5.1.5.1 Neural Network Classification ...........................................61
5.1.5.2 Fuzzy Classification............................................................63
5.2 Supervised Classification Using Multispectral Information .........................69
5.3 Exploring the Spatial Dimension...................................................................78
5.4 Using Contextual Information .......................................................................84
5.5 Taking Advantage of Morphology Parameters..............................................98
5.6 Taking Advantage of Texture.......................................................................100
5.7 Adding Temporal Dimension.......................................................................102


Chapter 6 Advanced Object Image Analysis....................................................109
6.1 Techniques to Control Image Segmentation within eCognition.................109
6.1.1 Using Ancillary GIS Layers to Contain Object Boundaries...........109
6.2 Techniques to Control Image Segmentation within eCognition.................115
6.2.1 Spatial Filtering................................................................................115
6.2.2 Principal Component Analysis (PCA).............................................116
6.2.3 Ratioing ............................................................................................119
6.2.4 Vegetation Indices ............................................................................119
6.2.4.1 Ratio Vegetation Index (RVI) ...........................................120
6.2.4.2 Normalized Difference Vegetation Index (NDVI) ...........120
6.2.4.3 Soil-Adjusted Vegetation Index (SAVI) ...........................121
6.2.5 RGB-to-HIS Transformation............................................................122
6.2.6 The Tassel Cap Transformation.......................................................122
6.3 Multiscale Approach for Image Analysis ....................................................130
6.4 Objects versus Spatial Resolution ...............................................................132
6.5 Exploring the Parent–Child Object Relationships.......................................134
6.6 Using Semantic Relationships .....................................................................137
6.7 Taking Advantage of Ancillary Data ...........................................................139


Chapter 7 Accuracy Assessment.......................................................................141
7.1 Sample Selection..........................................................................................141
7.2 Sampling Techniques ...................................................................................142
7.3 Ground Truth Collection..............................................................................142
7.4 Accuracy Assessment Measures ..................................................................143
References..............................................................................................................151
Index ......................................................................................................................155

Download Link

Multispectral Image Analysis Using the Object-Oriented Paradigm