Research On Algorithm Of Re-light Illumination Based On Single Face Image

One of the most important problems in face recognition is the illumination problem.When the illumination changes,the face image will change greatly.The same person face images taken in different light conditions is greater than the difference between different face images taken in the same light conditions shown in gray on the effect of illumination on the image,so the illumination problem has become one of the difficulties of face recognition.Now there are three ways to solve the problem of image illumination: The first is the transform based methods,namely linear or nonlinear transform in image spatial or frequency domain,to eliminate the effects of illumination on image.The second is based on illumination sample synthesis,which first estimate the need to identify the human face image the light conditions,and then simulate the standard illumination conditions,the last change the face image illumination.The third is based on the face model,Through reconstruct the 3D model of the face imageand and then it combine with the light cone or spherical harmonic,calculate the illumination conditions and relighting.Because the spherical harmonic model needs 3D face information,generally,a single face image based on 3D face model reconstruction,but the traditional construction algorithm of 3D face based on single image heavy computational complexity and time consuming.To solve this problem,a fast and robust method is proposed in this paper,which uses a single frontal face image as input to reconstruct a 3D face.The method can be divided into three steps: the first step is to extract the feature points of the human face.The second step is the adjustment of the model in the X-Y plane and the third step is model adjustment on Z-axis direction.Compared with the current methods,the experimental results show that the proposed method can reconstruct the 3D face model with real time.By using this method,we can get the normal vector of each point of the reconstructed 3D face model,then through the model of spherical harmonics,we can get the nine light coefficients related to illumination.By changing the nine illumination coefficients,we can generate the face image under different illumination conditions.