The Algorithm For Remote Sensing Image Segmentation Based On Modified Fuzzy C-means

Fuzzy cluster analysis of non-supervision of an important branch of pattern recognition. It has a wide range of applications in pattern recognition, data mining, computer vision and fuzzy control and some other fields. The most commonly used fuzzy clustering algorithm is Fuzzy C-means (FCM) clustering algorithm, which is a more effective combination of unsupervised clustering and the concept of fuzzy sets image segmentation techniques,But the fuzzy C-means (FCM) clustering algorithm has some defects, it is vulnerable to the impact of the initial cluster centers and the initial membership matrix may converge to local minimum value and effect the segmentation. In a large number of Case data set, such as remote sensing images, fuzzy C-means (FCM) will result in time-consuming iterative algorithm. To overcome these shortcomings, some research proposed genetic algorithm (GA), ant colony optimization (ACO) and particle swarm (PSO) algorithm combined with the FCM method. This paper presents two algorithms for clustering image segmentation: one is image segmentation algorithm combined with shuffled frog leaping and fuzzy C-means (FCM) .It's main idea is to use leapfrog algorithm (SFLA) and fuzzy C-means (FCM) algorithm to overcome the FCM algorithm which is vulnerable to the initial cluster centers and membership matrix of the initial segmentation result is not satisfactory leaving the defect. Experiments show that the algorithm which is compared with the recently proposed fuzzy C-means (FCM) and fuzzy PSO (FPSO) algorithm for image segmentation turns out to get a better image segmentation results. Another algorithm is modified fuzzy C-means (FCM) and Particle Swarm (PSO) combined with adaptive image segmentation algorithm. It's main idea is the introduction of the upper and lower cut set parameters to dynamically adjust the membership function, accelerate the convergence rat and introduce a number of adaptive clustering algorithm, so that the cluster segmentation of image adjustments can be adaptive number of clusters. Experiment shows that by comparing this algorithm with the standard fuzzy C-means (FCM) and particle swarm (PSO) algorithm, under the same split base, the number of clusters to optimize the image and make the convergence faster.