Image Segmentation Method Based On Genetic Algorithm

Recently, data of images has an explosive growth; images are becoming more and more important in our life. Facing huge data, it is a major task of image processing that how we extract useful information from the images. Image segmentation is a technology and procedure that decomposes image into some areas with features and extracts the objects that we are interested in. As a key technology of image processing, image segmentation has a direct effect of the quality of image analysis and image interpretation and image recognition.In recent years, the research of image segmentation is always the hot issue. Researchers have proposed thousands of algorithms of image segmentation. Fuzzy c-means clustering algorithm is a very popular and effective method to solve image segmentation. As higher requirement of the accurate of image segmentation, FCM algorithm also have progressed, And because of accuracy, FCM algorithm with local spatial information has become a hot spot. Particularly, FCM_S and FLICM algorithms are representative.According to the research of FLICM and the relevant algorithms, we found and proved that the solution of FLICM is not consistent with the minimizing of objective function and the frame of FLICM doesn’t have a closed-form solution, which leads to no convergence, so the same to the algorithms that use the frame of FLICM. We improved the objective function to absorb the local spatial information preferably. We also designed a program to solve the problem of function optimization, proposed a new image segmentation algorithm based on genetic algorithm (GAFLICM). The performance of GAFLICM on various synthetic and real images shows that GAFLICM algorithm is more insensitive to noise and enhance the accuracy of image segmentation.