A New Method of Color Tongue Image Segmentation Based on Random Walk

Abstract

This paper introduced a kind of new method for color tongue image segmentation by improving random walk algorithm. Firstly, we introduced an improved toboggan algorithm which adopted new classification rules to classify a image into initial regions. Secondly, we built a weighted-graph according to initial regions in which only those adjacent regions were connected. Thirdly, we adopted random walk algorithm to segment images by newly designed weight function. Fourthly, we used mathematical morphology operations to remove small holes on the target region of the segment result of the third step. In the experiment, we compared our method with traditional random walk algorithm. And as the experiment results show, our method achieves basically ideal effects, which are much better than those of traditional random walk image segmentation algorithm.

DOI: http://dx.doi.org/10.11591/telkomnika.v12i6.5390