Research And Implementation Of 3D Tongue Image-reconstruction Algorithm Based On Stereo Vision

Tongue diagnosis,also known as tongue inspection,is a method of detecting disease by observing the variation of patients' tongue body and tongue coating,belonging to the category of inspection diagnosis of traditional Chinese medicine,different from the diagnostic method of western medicine.Intelligent study based on tongue picture is becoming more and more widespread,due to the reason that tongue diagnosis has very important value in Chinese medicine.Most of the research is based on 2D image,judges the physiological condition of the patients via the color of the tongue coating.However,as the length and depth of the crack and the thickness of the tongue coating also play a supporting role in disease diagnosis,3D tongue image is urgently needed.This paper reconstructed the surface of the tongue in the way of stereo vision,elaborated on the theory and practice of stereo vision and reconstructed the tongue characteristics in a relatively good quality.The main work of the paper are as fellow:1 Studied the model and principle of the camera,wrote the capturing and calibrating program.2 Studied and summarized the usual stereo matching strategy and algorithm,put forward the stereo matching algorithm based on direction weight,realized local stereo matching in high speed,got the relatively precise dense disparity picture,.3 Constructed the software of tongue 3D reconstruction,integrated the function of distortion correction,tongue picture segmentation,epipolar rectification,stereo matching algorithm,Point cloud de-noising,point cloud data reduction,Point cloud triangulation and 3D demonstration.4 Studied the common tongue segmentation algorithm,developed the semi-automatic tongue acquiring algorithm based on the snake algorithm,reached a good effect in experiment.5 Constructed stereo reconstructed system,analyzed and tested the theoretical and practical precision of the system.