Image Segmentation Using Statistical Features and Block Processing

This paper proposes a new approach for image segmentation using Fuzzy C-mean classifier. The technique statistically analyses the data for image segmentation. In this approach, the image is initially subdivided into small blocks. Considering the average values of pixels in the blocks and their variances, adjacent blocks may be combined to create a larger block or a block may be further subdivided into smaller blocks. The size of the initial blocks in an image depends on the range of pixels’ variations so that for an image with a larger variation, blocks with a smaller size are required. Segmentation results of the proposed technique on a number of images indicate that it has a better performance compared to the existing approach.