An evolutionary algorithm for true-colored image quantization

Other Title(s)

الخوارزميات التقدمية لتكميم الصور ذات الألوان الحقيقية

Dissertant

Ulwan, Zaynah Ahmad

Thesis advisor

Atiyyah, Bara Ali

Comitee Members

al-Abbudi, Bushra Qasim
Abbud, Luayy Kazim

University

University of Baghdad

Faculty

College of Science

Department

Department of Computer Science

University Country

Iraq

Degree

Master

Degree Date

2004

English Abstract

The problem of color quantization is to represent full color RGB images, where each pixel is typically described by three 8-bit color samples, in an approximation fashion by a relatively small number of colors, thus the full color space, consisting of about 16 million colors (224 ) is represented by a small number of colors.

Historically, the number of colors used has been determined by the depth of the display frame buffer, often 8 bits (i.e., 256 colors).

The contribution of this work is to exploit the evolutionary idea of both Genetic Algorithms and Evolution Strategies for true color image quantization problem.

The components of an evolutionary algorithm are utilized to accommodate this quantization problem.

An EA individual can represents in its genotype level a potential palette for the true-colored image at hand.

Then, the phenotype of this EA individual represents the image but at quantized 256 or 16 levels.

Selection and Perturbation operators are also directed toward exploiting good palette portions of mating individuals to form new population.

Results of the implemented Genetic Algorithms and Evolution Strategies are sufficient with relatively small population size and few number of generation evolutions.

Main Subjects

Mathematics

Topics

• Algorithms
• Computer software
• Digital electronics

No. of Pages

61

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-605527