Two-Dimensional Entropy Image Segmentation Based On Genetic Algorithm

Image segmentation is one of the basic and important technique in computer vision. Image segmentation is the process and related methods to segment an image into different sub-images with different characters and derive some interested objects. It is a key step from image process to image analysis, and it plays an important role in image engineering. The quality of the segmentation will directly affect the performance of the vision system. So the study on the image segmentation is always one of the key points in the image technique study, many image segmentation algorithms are advanced, and it is widely used in computer vision, pattern recognition and medicine image processing.Threshold image segmentation method is in common used in the image segmentation. It distinguishes the image into objects and background by picking threshold, so the threshold picking is the key technique. The traditional segmentation method based on maximum histogram entropy can not obtain ideal result for lower signal noise-rate images. But 2D maximum entropy method can improve the anti-noise ability by fully using the information of gray distribution and special correlation of neighboring pixels. In this paper, a 2D maximum entropy method is expounded detailedly, and genetic algorithm is improved, and especially a new adapting mutation operator is proposed. The result of experiment shows that 2D maximum entropy method is of higher robustness, and the improved genetic algorithm can converge the optimum threshold quickly and has good optimum effect.This method is a complementarity to the image threshold segmentation method. The result of experiment shows, this algorithm is feasible and credible, giving an ideal image segmentation outcome, and more fast and steady than conventional method. It has great potential in the domain of image processing.