Orbit determination from three angles observations in the presence of perturbations

Other Title(s)

تحديد المدارات من ثلاث مشاهدات بوجود الاضطرابات

Dissertant

Zainal, Ahmad Qadir Izzat

Thesis advisor

al-Muhammadi, Abd al-Rahman Husayn Salih
al-Saati, Akram Yunis Thanun

University

University of Baghdad

Faculty

College of Science

Department

Astronomy and Space Department

University Country

Iraq

Degree

Ph.D.

Degree Date

2007

English Abstract

The classification process is an important task in digital image analysis.

In this research, an introduction was presented for the idea of using the concept of fractal dimension in the color textural images classification and discrimination processes.

Fractal parameters are good choice for modeling natural color textured surfaces and they have been recognized as important features in classifying images because the fractal dimension for surface correlates nearly perfectly with the perception of roughness in many situations.

The use of new set of fractal parameters to identify visual color textures was explored in this thesis.

The fractal dimension spectra were adopted for describing, discriminating and classifying different visual textures.

Two methods for estimating the fractal dimension spectra are proposed; the first method is the modified binarization method and the second is the traditional box counting method.

Also, a focus on these fractal dimension spectra, with their capability and recognition accuracy for a set of natural color textures, were evaluated.

The problem of texture images classification was presented, and two techniques based on using the fractal dimension spectra were derived.

The first technique is based on supervised classification technique that is based on the minimum distance measure classifier.

While, the second technique is a non supervised using the k-means algorithm.

In this research, a testing procedure was conducted to evaluate the degree of sensitivity for each kind of fractal dimension spectra to the visual variation of the different textures.

This procedure has shown satisfactory results, which can satisfy the usage of fractal dimension spectra as textural discriminating criteria.

The proposed classification methods were applied on two important set of images; the first set consists of remote sensing images, and the second set consists of breast tumors images.

The higher recognition accuracy in the first set of textures images was (100%), while in the second set was (98.8%).

The classification and discrimination results were presented to demonstrate the efficiency of these proposed fractal based techniques.

The experimental results lead to the conclusion that the fractal dimension spectra provide substantial tools to discriminate textures and, hencefore, they are useful for the discrimination and classification of color textures.

Main Subjects

Astronomy

No. of Pages

153

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-604467