SVR based color calibration for tongue image

Abstract

Tongue color is one of the most important pathological features for Computer Aided Tongue Diagnosis System (CATDS). Color distortion of tongue image, which is often caused by the inconstancy of lighting conditions, can seriously affect the validity of diagnosis results obtained by CATDS and furthermore, impair the interchangeability among tongue images captured by different devices. To circumvent the above problems, we present a novel color calibration model based on Support Vector Regression (SVR) method in this paper. First, a new colorchecker which provides a suitable reference for the following color calibration is produced by color clustering of 322 tongue images. Then, a comparative analysis between the SVR based color calibration method and Polynomial Regression based one is carried out. Results indicate that SVR based color calibration model, cooperating with the proposed colorchecker, provides an outstanding calibration effect and plays an important part in the image acquisition of CATDS.