3D Face Reconstruction Algorithms Based On Features Constraints

3D face reconstruction is a hot research topic with great challenge, involves many other fields such as applied mathematics, image processing and machine vision and has potential applications in biometrics recognition and3D animation. With the society informationization and the development of computer technology,3D face reconstruction will receive deeper development.3D face data is the basis of face reconstruction. The way that get3D data base on binocular stereo vision has received widely attention because of its advantages such as low-cost, friendly user interface and so on. This article discusses the method by researching on algorithms of camera calibration, stereo matching and point cloud triangulation. The main contributions of this article are summarized as follow:Firstly, by analyzing the gray information of black-white checkerboard image, a method that combined with Harris operator and BW operator is used for corner detection and extracting sub-pixel corner points’ coordinates that used for camera calibration to obtain precise cameras’ intrinsic and extrinsic parameters.Secondly, with the advantages of feature matching and region matching, an improved block matching algorithm based on the constraints of facial features is presented. In this algorithm, human face has been segmented into several regions according to the results of Adaboost face detection and different matching strategies are adopted in different regions to obtain dense and sub-pixel disparity map of human face.Thirdly, the improved experiments and algorithms of this paper were implemented on MATLAB and then the results are analyzed briefly.Finally, summarizing the problems of this paper, and looking ahead future research works that need to be improved in the next step.