This paper proposed a new method using statistic feature to identify if a tongue is a cracked tongue, which is one of the most frequently-used visible features on diagnosis of traditional Chinese Medicine. We first detect the wide line in the tongue image, and then extract statistic feature such as Max-distance of detected area, the ratio between Max-distance and size of detected area and so on. We train a binary SVM based on these statistic features to build a classifier for cracked tongue. An experiment based on the proposed scheme has been carried out, using 196 samples of cracked tongues and 245 samples of non-cracked tongues. The results of the experiment illustrate that the recognition accuracy of the proposed method is more than 95%.
[1]
David Zhang,et al.
Extracting Tongue Cracks Using the Wide Line Detector
,
2008,
ICMB.
[2]
Nobuyuki Otsu,et al.
ATlreshold Selection Method fromGray-Level Histograms
,
1979
.
[3]
N. Otsu.
A threshold selection method from gray level histograms
,
1979
.
[4]
David Zhang,et al.
Detecting Wide Lines Using Isotropic Nonlinear Filtering
,
2007,
IEEE Transactions on Image Processing.
[5]
David Zhang,et al.
Computerized tongue diagnosis based on Bayesian networks
,
2004,
IEEE Transactions on Biomedical Engineering.