Research On Image Based 3D Human Face Automatic Generation And Editing Algorithm

Since 1970 s,3D face modeling has been one of the hottest research topic in computer graphic.3D face modeling technology has been widely applied in Business.In the field of animation,movie and computer game,many virtual characters are created using 3D face modeling technology.In the field of medicine,3D face modeling technology can be used to reconstruct patient's head and can be used in virtual surgery.In the field of biometric identification,3D face modeling technology can be applied to face identification.This technology can also be used in many other fields like virtual reality,human-computer interaction and network aided instruction.Therefore,research on 3D face modeling is a meaningful and valuable work.Currently image based 3D face reconstruction algorithm can produce a face model which is roughly fitted the face in image.However there are obvious errors in the reconstruction of the detailed part of the face.Besides,the reconstruction of the face texture is not precise.The limitations of these algorithms are as followed.No optimization for the reconstruction of the detailed part of the face.Facial features extracted from the images are not complete and accurate.The 3DMM face color texture model is not capable of reconstructing detailed facial texture.We make improvements for these limitations.In the reconstruction algorithm,facial feature points and optimization for this part are added.We improve the edge extraction algorithm to make the extracted edges more complete.Finally,We generate the texture image of the 3D face model through aligning and merging facial images.Our algorithm detects human face using Haar-like features face detection algorithm,calculate positions of facial feature points using face alignment algorithm,and then extracts facial edges using edge extraction algorithm.After extracting facial features from the image,we align the corresponding feature points,align the facial edges and fit the pixels intensity between the 3D face model and the face image.Finally,we align and merge the frontal view face image and the side view face image to generate the complete facial texture and process a texture mapping on the 3D face model.Besides,we optimizes the lighting parameters according to the Phong lighting model.The reconstructed face model can be represented by alow-dimensional parameter vector and can be edited through adjusting the parameter value.This paper proposed a 3D human face automatic generation and editing algorithm which works in any device that has a camera.It can be directly applied to movie,animation and computer game production.It simplifies the process of 3D face modeling.It can also be used in 3D face recognition and other face relative software development.