An Ant Colony Algorithm For Fuzzy Clustering In Image Segmentation

Image segmentation is a technology image processing, which refers to the image into various characteristics of regional for extracting useful target, it is a key step from the image segmentation to the image analysis, and it plays a very important position in the image project, image segmentation shows actual application in many fields, such as computer vision and pattern recognition, coding, and medical image analysis.The method of artificial life helps people understand biology, and in the robot, computer graphics is applied successfully, etc. The ant colony algorithm of artificial life as a branch of the algorithm is true of ant colony activities and gradually developed a kind of simulation ant colony algorithm of intelligent behavior. It has strong stability and distributed computing and combined with other algorithm, advantages of individual information exchange and constantly is finding better solutions.According to the image itself exists many uncertainties and imprecise, people found that the fuzzy theory describing uncertainty of image is very good, but the image segmentation problem is exactly the pixel image classification problem, some scholars in recent years apply image clustering in image segmentation, the effect is better than the traditional image segmentation method, but the classical fuzzy clustering method, some problems still exist.This thesis based on fuzzy clustering image segmentation method for the research focus in detail, analyzes its principle and development situation, and completes the following improvement to the deficiency of the clustering algorithm for image segmentation:①Introducing ant, using the global and robustness of ant colony algorithm can effectively overcome the fuzzy clustering of initial parameters of sensitive, the ant cluster of sensitive initial clustering algorithm, By gathering ant algorithm that initial cluster number and the center, as the initial parameters of fuzzy clustering, thus the image segmentation. The improved algorithm reduces the segmentation of sensitive degree to initial parameter.②This paper introduces and analyses the fuzzy clustering algorithm and its development, through a large number of literature and the experiments analysis the influence of fuzzy weighted factor m to the fuzzy clustering, when fuzzy weighted factor 1≤m≤3, the fuzzy clustering image segmentation have a best effect. ③According to the traditional fuzzy clustering algorithm poor robustness, Combining with the Markov airport proposed an improved fuzzy clustering image segmentation algorithm. Fuzzy clustering method does not consider nuclei pixels and neighborhood, and other correlation of pixels are very sensitive to noise. Using fuzzy Markov space constraints on fuzzy clustering algorithm, can exert space constraints efficiency enhancement algorithm robustness. For different kinds of image segmentation experiment shows that this algorithm is improved significantly lower than before the mistake.