Image Segmentation Based On Genetic Algorithm

Genetic Algorithm (GA) is a kind of iterative intelligent optimization algorithm. The character of GA is robust, parallel, adaptive and fast convergence speed, which is used to get the segment threshold in the application of image segmentation.Firstly, GA is introduced in detail, including the biology basis of GA, development history, basic principle, character and simple work flow.Secondly, some typical methods of image segmentation are presented, including threshold value, brim detection, area tracing, and coordinate position and so on. And image segmentation methods combined special theoretic tools are introduced, including neural networks, wavelet analysis and transform, mathematical morphology, fuzzy mathematics, partial differential equation and so on. The evaluation methods to the image segmentation algorithms are introduced. These methods can not only improve the performance of the algorithm but also have the sense of researching the fresh technique.In this paper, the improved GA is proposed to reducing the chances fall into the local minima and enhance the global search ability. In the proposed method, the cluster degree of the population is calculated through the evaluation procedure. Therefore, the algorithm can adjust the mutation strategy according to the cluster degree. If the cluster degree is higher than a preconcerted value, the mutation rate will be set to a big value; and if the cluster degree is lower than a preconcerted value, the mutation rate will be set to a small value. The improved GA and simple GA are used segment two different images based on the best entropy double threshold method. The results of emulation indicate that the improved GA can get better' segmentation results than the simple GA.Finally, some conclusions are presented.