Research On Fuzzy Clustering Algorithm For Image Segmentation Based On Spatial Information

Image segmentation is just to divide an image into different sub-images with different characters and extract some interested objects. It is the most essential and important content of research on low-lever computer vision, and is a key technique for image analysis, understanding and description because the quality of segmentation results affects the quality of succeeding analysis, recognition and explaining. Image segmentation is applied in a lot of fields such as computer vision, image coding, pattern recognition, medical image and so on.Images themselves are very uncertain and inaccurate. It is found that fuzzy theory is able to give a good description of such uncertainties and image segmentation is just the classification of image pixels. In recent years, some experts are making efforts to apply the fuzzy clustering method in image segmentation, and it is more effective than the traditional image processing method. However, there are still some problems with classical image segmentation based on fuzzy clustering, and the main problem is that segmentation result is very sensitive to noises. The analysis indicates that one of the reasons is inadequate use of the image space information. This paper focuses on the research on how to reasonably utilize the space information, and consists of the following main contents:1) Standard FCM algorithm for image segmentation is discussed deeply, and many problems of the algorithm are studied, such as initialization of the number of clustering and class centroid, setting of weight exponent and so on.2) Research on FCM algorithm for image segmentation based on two-dimensional feature of gray and spatial information is done and it is indicated that the weight of gray and spatial information should not be fixed but determined by information of each image itself. FCM algorithm for image segmentation based on gray and spatial information feature weighted is proposed, which optimizes the weight of spatial information by evolutionary strategy utilizing clustering validity function as objective function. Then the algorithm is applied in synthetic test image and realistic image, and segmentation results are compared with that of standard FCM algorithm and FCM algorithm based on two-dimensional feature, which suggests that proposed algorithm has better effect.3) Another approach incorporating spatial information is researched, which...