Implementation of image segmentation and reconstruction using genetic algorithms

This paper describes an approach of using genetic algorithms (GAs) for image segmentation. The approach proposes to solve the problem based on the idea of using genetic programming to discover effective problem-specific filters capable of highly and selectively emphasizing some characteristics of the image. Genetic algorithms work on the principle of simulation of the evolution of individual structures via processes such as selection, mutation, and reproduction. Implementation of image segmentation using GAs involves identifying a suitable binary coding strategy, defining a fitness evaluation function, designing a population´ (set of chromosomes), defining genetically inspired operators such as crossover and mutation to evolve new population and deciding the termination of the evolutionary search for the optimal solution. Segmentation is a non polynomial type of problem. This paper describes the use of genetic algorithms in the randomized search of solution to the segmentation problem, with the initial population built using the state space techniques and then evolved using genetic operators and the fitness function. Accuracy of the results is found to vary with the fitness function. The result is got in the binary form due the coding strategy adapted. A mask is then implemented to extract and reconstruct the segmented image from the original image.