A Snake-Based Approach to Automated Segmentation of Tongue Image Using Polar Edge Detector
Hongzhi Zhang,Wangmeng Zuo,Kuanquan Wang,David Zhang
DOI: https://doi.org/10.1002/ima.20075
IF: 2.177
2006-01-01
International Journal of Imaging Systems and Technology
Abstract:Tongue diagnosis, one of the most important diagnosis methods of Traditional Chinese Medicine, is very competitive as a candidate of remote diagnosis method because of its simplicity and noninvasiveness. Recently, considerable research interests have been given to the development of automated tongue segmentation technologies, which is difficult due to the complexity of pathological tongue, variance of tongue shape, and interference of the lips. In this paper, we propose a novel automated tongue segmentation method via combining polar edge detector and active contour model (ACM) technique. First, a polar edge detector is presented to effectively extract the edge of the tongue body. Then we design an edge filtering scheme to avoid the adverse interference from the nontongue boundary. After edge filtering, a local adaptive edge bi-thresholding algorithm is introduced to perform the edge binarization. Finally, a heuristic initialization and an ACM are proposed to segment the tongue body from the image. The experimental results demonstrate that the proposed method can segment the tongue body accurately and effectively. A quantitative evaluation on 200 images indicates that the normalized mean distance to the closest point is 0.48%, and the average true positive percent of our method is 97.1%. (c) 2006 Wiley Periodicals, Inc.