Degree Type

Thesis

Date of Award

1998

Degree Name

Master of Science

Department

Electrical and Computer Engineering

Abstract

Image segmentation is a fundamental component of picture processing and image analysis. Segmentation of an image entails the division or separation of the image into regions of similar attributes. The most basic attribute for segmentation is the image intensity (luminance for a monochromatic image). Several classical methods for image segmentation exist and it is well known that these methods are more or less heuristic and specific to a particular application.

Genetic Algorithms (GA) are stochastic search methods, the functioning of which is inspired by laws of genetics, natural selection and evolution of organisms. Their main attractive characteristic is the ability to deal with hard combinatorial search problems efficiently, where parallel exploration of the search space, eliminates to a large extent the possibility of getting stuck in the local extrema. The basis of the theory is that individuals tend to pass on their traits to their offspring and the fittest of the individuals tend to have more offsprings. In effect, the tendency is to drive the population towards favorable traits. Over long periods of time, entirely new species are produced which are better adapted to a particular ecological condition.

This thesis proposes a simple and robust method for image segmentation that is based on the application of Genetic Algorithm and Mathematical Morphology. The image is divided into nonoverlapping subimages and the genetic algorithm is applied to each subimage, starting with initial random populations. Each individual of the population is evaluated using an appropriate fitness function. The best-fit individuals are selected and mated to produce offsprings to form the next generation. Morphological operations are used to produce the next generation along with the crossover and mutation operators. The algorithm converges to yield the final segmented subimage. These segmented subimages then are combined to form the final result. The feasibility of applying genetic algorithm and morphological operations to an image segmentation problem is evaluated and results are presented and discussed.

DOI

https://doi.org/10.31274/rtd-180813-6121

Publisher

Digital Repository @ Iowa State University, http://lib.dr.iastate.edu

Copyright Owner

Ming Yu

Language

en

Date Available

2013-12-13

File Format

application/pdf

File Size

106 pages

Share

COinS