Anisotropic Diffusion with Morphological Reconstruction and Automatic Seeded Region Growing for Color Image Segmentation

A watershed color image segmentation algorithm by integrating anisotropic diffusion with morphological reconstruction and automatic seeded region growing is proposed in this paper, the anisotropic diffusion with morphological is used for image smoothing, which can maintain image edges with noises eliminated, as a result, the watershed´s over-segmentation is decreased effectively. And then, an automatic seeded region growing algorithm based on the regions similarity of mean hue value and mean saturation value of the watershed segmentation result is developed. In order to reduce the number of small regions, a small region removing algorithm is also designed. The experimental results have demonstrated that algorithms can obtain better segmentation results compared with traditional algorithms.