Authors : S. Kothainachiar and R.S.D. Wahita Banu
Abstract: A Supervised Watershed Algorithm segments the image into as many regions as the number of markers, interactively placed by the user in the Region of Interest (ROI). Image Smoothing acts as the pre-processing step for segmentation. A non-linear Partial Differential Equation technique (PDE) or scale-space diffusion technique is used for image smoothing. This algorithm simplifies the problem of initialization as the image pixels are sorted previously according to difference between the maximum and minimum values in neighbors of the current pixel. This sorting makes an order of processing and ensures that the pixels around the edges have their processes postponed. The local seed points are automatically detected and labeled. To avoid over-segmentation problem, this algorithm merges the non-significant regions based on the markers. The user can repeat the process, until satisfactory results are obtained. Accurate segmentation results are obtained for gray scale images and color images by using this simple algorithm.
S. Kothainachiar and R.S.D. Wahita Banu , 2006. A Supervised Watershed Algorithm for Image Segmentation. Asian Journal of Information Technology, 5: 1441-1446.