Robust tongue segmentation by fusing region-based and edge-based approaches

Abstract

A tongue diagnosis system can offer significant information for health condition. To ensure the feasibility and reliability of tongue diagnosis, a robust and accurate tongue segmentation method is a prerequisite. However, both of the common segmentation methods (edge-based or region-based) have respective limitations so that satisfactory results especially for medical use are often out of reach. In this paper, we proposed a robust tongue segmentation method by fusing region-based and edge-based approaches. Before segmentation, ROI (region of interest), which will be used as input for the subsequent segmentation, was extracted by a novel way. Next, we merged adjacent regions utilizing the histogram-based color similarity criterion to get a rough tongue contour. It is essentially a region-based method and hence the results are less sensitive to cracks and fissures on surface of the tongue. Then, we adopted a fast marching method to connect four detected reliable points together to get a close curve, which is based on edge features. Contour obtained by region-based approach was utilized to act as a mask during fast marching process (edge-based) and the mask added limits so that the ultimate contour will be more robust. Qualitative and quantitative comparisons show that the proposed method is superior to the other methods for the segmentation of tongue body in terms of robustness and accuracy.