Application Of FCM Clustering Based On Harmony Search Algorithm 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 image coding, pattern recognition, computer vision and medical image analysis.According to the image itself exists many imprecise and uncertainties, 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, in recent years, some scholars 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 C-means 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:First of all, this paper describes the harmony search (HS) algorithm, which is a heuristic algorithm. But the harmony search (HS) algorithm traps into local optima easily, and it has low convergence accuracy when it is used for the optimization of complex functions. In order to overcome these shortcomings, an improved harmony search (IHS) algorithm is proposed. Through test function to compare two algorithms, it is improved that harmony search algorithm has better features.And then, introducing the harmony search (HS) algorithm and the improved HS (IHS) algorithm, using the robustness and global of two algorithms can effectively overcome the fuzzy clustering of initial parameters of sensitive, the harmony search (HS) cluster of sensitive initial clustering algorithm, By gathering the harmony search (HS) algorithm that initial cluster number and the center, as the initial parameters of fuzzy clustering, thus the image segmentation.Finally, This paper carries on the image segmentation by using cluster ant clustering algorithm, the original harmony search clustering algorithm and improved harmony search algorithm respectively, and the results show that the treatment based on improved harmony search FCM algorithm has high accuracy and high segmentation speed, as well as get the real image clustering number, and it can also reduce the sensitive degree of initial parameters.