Three-dimensional Face Reconstruction Research Based On Varying Illuminations

3d(Three-dimensional)face reconstruction can restore the geometric structure of face through 2d images.Compared with 2d images,the 3d face model has an intrinsic pose invariance.It won't be influenced by the change of environment and observation angle,which can be widely used in face recognition,biomedicine and other fields However,due to the acquisition equipment and reconstruction algorithm,this method mainly have two problems:One is that the reconstruction accuracy is insufficient which is reflected in the loss of high-frequency details such as facial wrinkles and pores.And the other is the loss of specular map,which results in the poor light reflectionTo solve the above problems,through theoretical analysis and experimental verification,this paper focused on how to restore the high frequency details and calculate the specular map.The main research works and innovations include:1.Design and set up the multi-view image acquisition equipment with varying illuminations.Compared with traditional equipment,it can provide a variety of different lighting conditions by adjusting the arrangement and combination of light sources.And it's able to catch the photos from multiple perspectives at the same time2.Propose a new kind of high precision reconstruction algorithm which can effectively restore the high frequency details smaller than one sub millimeter.This paper combined the multi-view method,photometric stereo and special multichannel filtering algorithm to extract the display map and finally restore the 3d face model with highly accuracy through images with different lighting conditions.The loss of high-frequency details is solved and the time cost is reduced.3.Propose a fast method to extract the specular map by the polarized light.This paper improved the whole acquisition process by linear polarizer to get the missing specular map in less than 1 second.The poor light reflection is solved.