Dissertation

Comparison pixel-based and object oriented classification for Informal settlements using SPOT-5 imagery

By



Abstract

In this study we try to explore the concept of Comparison pixel-based and object oriented classification for Informal settlements using SPOT-5 imagery. The main focus of the research is on remote sensing techniques and its relation with the urban features. The research also analyzes many aspects of the identified topic and intends to explore the informal settlements in Abu Arish city. Finally the research describes various factors which are responsible for the procedures which tend to enable the transfer of knowledge. Moreover, this study also focuses on the rise of informal settlements in many regions of the Kingdom of Saudi Arabia. For example, we find a high density of informal settlements areas in the province of Jeddah and in the holy city of Mecca. The informal settlements in the western and central regions of Saudi Arabia are already undergoing amelioration through appropriate plans to resolve the problem. This study tends to take into fold the tools such as documentation, Quality assessment, classification and pre-processing to under the different classification techniques and underlying notions. During the course of the study Two SPOT-5 images will be used in order to understand the nomenclature of the identified topic. It would be safe to state the fact this study tends to provide a clear picture regarding the limitations of the pixel based and object oriented method to meet the pretext of the paper.

Table of Contents

ABSTRACTII

CHAPTER 1: INTRODUCTION1

Aims and Objectives2

Study Area3

Methodology4

Pre-processing4

Classification5

Pixel-Based Classification5

Object-oriented Classification5

Material6

CHAPTER 2: LITERATURE REVIEW7

“Object-Oriented” Classification Techniques7

Image Segmentation Phase8

“Object-Oriented” Classification Phase10

Multi Resolution Segmentation Technique11

CHAPTER 3: CLASSIFICATION12

Pixel-based Classification12

Advantages of Unsupervised Classification13

Disadvantages of Unsupervised Classification13

CHAPTER 4: OBJECT-OREINTED20

Images Segmentation Phase20

Classification Technique21

Validation of Classification26

CHAPTER 5: QUALITY ASSESSMENT30

Classification Accuracy30

Project Time31

CONCLUSION32

CHAPTER 1: INTRODUCTION

In 2003, almost one billion people or 32% of the world's population lived in informal settlements in the developing countries. More recently, more than half of the population of the earth has been identified as living in informal residences in urban areas. These informal residential buildings were the result of many causes, for example some cities lack urban planning and are facing rapid growth in population. The 2003 United Nations Human Settlements Program pointed out that 90% of future population growth would be in cities and most of this growth would exist in the developing countries. In fact, unplanned rapid growth threatens both society and the environment in these cities. One problem is the absence of infrastructure in the cities, especially in informal settlements, resulting in unhealthy living conditions (Andrea & Eva, 2005).

Remote sensing techniques are gaining ground as satellite imagery becomes the most important data source to study urban change. Plus, it tends to prove significant in providing information on land cover for subsequent classification or analysis. Remote sensing techniques now provide information of high resolution which is regularly updated. Large collections of remote sensing imagery have provided a solid foundation for spatial-temporal analysis of the environment and the impact of human activities (Zhou et al., 2004). These data provide a 'freeze-frame' view of the ...