Skip to Main Content
According to the theory of traditional Chinese medicine (TCM), the tongue's color, texture and shape can reflect a person's physical health. Therefore, the tongue image analysis and matching are playing more and more important role in the tongue image's computer auxiliary diagnosis in TCM. Currently, the majority of the existing tongue image analysis methods are based on the single image feature (such as color, texture and shape), it is difficult to describe completely the characteristics of the tongue. This paper proposes a multi-feature fusion method (MFF) to improve the recognition rate for tongue image matching and tongue diagnosis in TCM. The optimization objective of this method is to minimize the number of the discordant pairs (inversions) between the predicted rank and the target rank, and then the proposed method learns a group of rational parameters on training data set to fuse multiple tongue image features. The experimental results show that the MFF method has very promising performance on tongue images matching and tongue diagnosis.