Segmentation of MELK images based on watershed algorithm
基于分水岭算法的MELK图像分割

This paper proposed a segmentation method for MELK images based on a watershed algorithm, which was applied to segment the MELK images that include little information. Employed several techniques to solve the over-segmentation problem in the watershed algorithm. Firstly, used a Gaussian lowpass filter to enhance the MELK images. Provided weighted difference operators to obtain the gradient images, and processed the gradient images by a watershed algorithm after thresholded them by the Ostu method. At last, segmented MELK images based on a region merging algorithm called RAG method. The experimental results indicate that the proposed method is feasible to segment the MELK images which include little information.