Research On Pixel Clustering Based Image Segmentation Algorithm

Along with the increase of image information content, image processing becomes one of the hottest research topics. Image segmentation is an important step in image processing technology. Segmentation results affect following image analysis and image understanding. Image pixel based clustering algorithm is used frequently in image segmentation. On account of image is affected by noise and other disturbance factors in the procedure of image acquirement, so image has the higher characteristic of fuzziness. Fuzzy clustering based image segmentation algorithm can process the blurry images, but this kind of method has its limitation. The former researches show that image segmentation algorithm based on type-2 fuzzy clustering is better to deal with the problem of image fuzziness. Therefore, two image segmentation algorithms, Linear Stretching-FCM (LS-FCM) and Membership Stretching-FCM (MS-FCM), based on type-2 fuzzy clustering theory are proposed in this paper. The two methods are based on the fuzzy c-means algorithm through enlarging the large membership (the possible target) and decreasing the small membership (the possible noise) for increasing their difference in order to get more accurate segmentation results. Experimental results show that the proposed algorithms have higher image segmentation accuracy and better noise immune ability. Moreover, MS-FCM algorithm has better segmentation performance and higher running speed.