Abstract: With the further development of modern tongue diagnosis, reasonable use of tongue 3D color point cloud data has become a key step in TCM diagnosis and treatment of various diseases and obtaining the objective and quantitative information. By combining modern 3D point cloud processing technology with traditional TCM diagnosis experience, an algorithm for the Euclidean cluster segmentation of tongue based on the Fast Point Feature Color Histogram(FPFCH) eigenvalue and the region segmentation based on normal of tongue crack extraction was proposed. The FPFCH eigenvalues consisted of the extended Fast Point Feature Histogram(FPFH) component and the Hue(H) color component as the discriminant condition of the tongue point cloud after the Euclidean cluster segmentation. The area was segmented by judging the threshold of the normal line angle and the point cloud of the tongue crack was extracted. A large number of experiments show that the algorithm can effectively complete the tongue segmentation and tongue crack extraction, which provides technical support for the research of object diagnosis of the tongue.