Tongue Texture Analysis Using Gabor Wavelet Opponent Colour Features for Tongue Diagnosis in Traditional Chinese Medicine

This article reports an application of using Gabor Wavelet Opponent Colour Features (GWOCF) for tongue diagnosis in Traditional Chinese Medicine (TCM). This project focuses on the standardization of the tongue diagnosis in TCM. The objective is to develop a set of quantitative measurement for tongue diagnosis using computer vision techniques. To achieve this goal, we propose to use GWOCF for determining different types of tongue proper and tongue coating textures from a tongue image. Two experiments are presented. In the first experiment, GWOCF is directly extracted from tongue image for recognition. In the second experiment, we employ colour information to pre-classify the known texture image before extracting GWOCF. 63 tongue images captured from 63 patients in Guangzhou University of Traditional Chinese Medicine's hospital are used for testing. It is found that with colour pre-classification process, 89% of recognition accuracy can be achieved. This is a revised version of our conference paper in [12].