Research Of The3D Face Modeling And Face Gesture Recognition And Its Application In Road Test System

In the field of research in computer vision, how to use the computer to locate anddescribe the interesting object and the corresponding study is a very importantresearch content. Face, as the most direct and important carrier for emotionalexpression and mutual exchange in our daily life and effective means to distinguishbetween individuals naturally becomes an important object of research in computervision. Face-related applications are also very widespread. Among them, RTS3,representing the Road Test System, using face image processing technology toanalyze the candidate’s head gesture and then determine whether the head movementsare illegal. While the current Road Test Systems only use2D face information toanalyze candidate’s head movements, which comes from the projection ofthree-dimensional face and is just the partial information of the face. There will bemany limitations when just uses2D face information to solve the problem of facegesture recognition.In order to improve the ability of analyzing head movements of candidates inRTS3, the thesis pays attention to the3D face information to analyze the face gesture.Because the3D face gesture recognition uses the intrinsic information of face thatcontains the spatial information, it can use much more information to analyze the facegesture than the face gesture recognition based on2D information.First of all, the thesis uses ASM to extract facial feature point from a singlepositive target face image without shelter, and then uses the information to modifyand adjust the lightweight mesh generic3D facial model Candide-3to generate acorresponding specific3D facial model. Then extract the texture information of thetarget face from the image, texture mapping on the adjusted specific3D facial modelto generate a realistic3D facial model, and then uses the specific3D facial model torecognize and simulate the face gesture. Finally, introduces the face gesturerecognition method based on specific3D face model into the RTS3to analyze andjudge the head movement of candidates.