Grid-Based Genetic Algorithm Approach to Colour Image Segmentation

Introduction

Image segmentation is a complex problem that has received lots of research in the past. Although there are many segmentation techniques, they all have their weaknesses. There is no single most effective technique that works best for every scenario.

Colour images increase the amount of information that needs to be processed, hence increasing the uncertainty. With such high levels of uncertainty and a problem that has no well-defined optimal solution, genetic algorithms have been used to a limited extent in segmenting images.

Segmentation is a computationally expensive task and the use of genetic algorithms to solve the problem requires large amounts of processing power. One way of obtaining this processing power is to make use of Grid computing. While genetic algorithm models have been developed and proven very successful for parallel architectures, almost none of this knowledge has been applied in developing a model for the Grid.

Flowers segmented with genetic algorithm

Goals

This research has the following major goals:

  1. Research existing image segmentation techniques and make a selection to experiment with.
  2. Develop a genetic algorithm solution to image segmentation using the knowledge from the experimentation in the first goal.
  3. Develop a Grid-based genetic algorithm model with the aim to use the processing power available on a Grid to increase the speed of the genetic algorithm image segmentation solution.
© Marco Gallotta and Keri Woods 2007