Reconstruction of 3D facial image using a single 2D image

In many scenarios related to image processing, 3D face reconstruction is considered an essential part. A robust and fast method of 3D face reconstruction from a single 2D image is presented. This method consists of three main steps including extraction of features from single image, calculating the depth of image and adjustment of a 3D model on the direction of Z-axis. In the first step, the features of image are extracted by using supervised descent method (SDM). Using SDM, face regions like facial components (eyes, nose lips) and face contours are detected. Second step consists of depth prediction by implementation of multivariate Gaussian distribution. Finally, 3D face is constructed with the help of features and the depth information and 3D database. The proposed method has been verified by conducting several experiments depicted in evaluation section. Our method is robust in nature and gives good results even using a single image, comparing to other methods that use multiple images for reconstruction of 3D images.