Evaluation of the digital image classification techniques

Dissertant

Muhammad, Sami Muhammad Ali Fadil

Thesis advisor

al-Sinnari, Adil Muhammad Ahmad

Comitee Members

Bilal, Khalid Hamid
al-Jund, Ammar Babakir

University

Omdurman Islamic University

Faculty

Faculty of Sciences and Technology

Department

Department of Information Systems

University Country

Sudan

Degree

Master

Degree Date

2011

English Abstract

Recently the digital image processing become so important because of use it in many purposes in our life.

The main Aim of this study is differentiate on between the various methods of classifying the Digital image.

This study is aim to evaluation and comparison between two methods of classification.

(Supervised classification of unsupervised classification).

The test done in this study has there phases.

First Phase the study depend on test's data include5 Bands and it's done by German satellite (RapidEye) for processing digital image they use special systems like.

(Erdas Imagine 9.2).

Data test have been explored and also Image contents have been improved and geometric correction.

Second phase is under go supervising classifying operation by using the maximum likelihood, Mahalanobis Distance, Minimum Distance and unsupervised classification.

The third phase is the most important phase of the test where we compare between the output of supervised classification and unsupervised classification images.

These studies provide that, the supervised classification present good results.

The supervised classification method is classifying all classes in a way identical to visual interview which done.

In example, 4.66% of study area is classified as Water opposite to unsupervised classification method where the percentage in the results of classification was reach 14.42 %.

The study has many recommendations and the most important recommendations are.

Use large number of bands especially in thermal purpose (x-ray).

The complete knowledge about the nature of study area for the purpose of improving supervised classification.

The complete depend on electro magnetic when determining the upper limit for classes when making unsupervised classification.

Check and test the various classification methods.

Main Subjects

Electronic engineering

Topics

• Evaluation
• Remote sensing
• Digital electronics

No. of Pages

53

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-364767