Research On Characteristics Processing Algorithm Of 3D Color Point Cloud

Tongue diagnosis is core content of tradition Chinese medical inspection,occupies an important position in the theoretical system of traditional Chinese medicine and clinical diagnosis,with the development of the modernization of tongue diagnosis,reasonable use of the tongue color 3D point cloud of TCM diagnosis has become the key to treat various diseases and obtain the objective quantitative information.By combining modern 3D point cloud processing technology with traditional TCM diagnosis experience,an imaging system of color 3D tongue diagnosis based on structured-light scanning measurement method has developed,and puts the algorithm of tongue segmentation and crack extraction based on point cloud,overcomes the disadvantages of the tongue image can't reflect the tongue contour and the length of crack.The major research contents are composed of the following three parts:1.An imaging system of color 3D tongue diagnosis based on structured-light scanning measurement method has developed,Through the point cloud acquisition experiment of different subjects,the system can accurately obtain the point cloud data of surface of tongue,and the color information matching is accurate and complete.2.Through the analysis and comparison of the existing point cloud segmentation and recognition algorithm,an algorithm for Euclidean cluster segmentation of tongue based on the FPFCH eigenvalue and region segmentation based on normal of tongue crack extraction are proposed.The FPFCH consist of the FPFH and the H component.3.Through the analysis and comparison of the existing point cloud filtering algorithm,using moving least squares method to process the tongue point cloud data,an algorithm for extraction of tongue crack based on the fusion of multi segmentation method.The region segmentation based on normal is by distinguishing the angle threshold to eliminate smooth point cloud,and get the tongue crack point cloud based on Euclidean clustering method.Experiments of tongue point cloud processing of different subjects show the algorithms of tongue segmentation and crack extraction can accurately complete the task,which has important significance for the Intellectualization of tongue diagnosis.