Fuzzy based seeded region growing for image segmentation

This study proposes a novel seeded region growing based image segmentation method for both color and gray level images. The proposed fuzzy edge detection method, that only detects the connected edge, is used with fuzzy image pixel similarity to automatically select the initial seeds. The fuzzy distance is used to determine the difference between the pixel and region in the consequent regions growing, in which the conventional regions growing is modified to ensure that the pixel on the edge is processed later than other pixels, and the difference between two regions in the regions merging. In the simulations, the proposed method outperforms other existing segmentation methods.