METHOD, DEVICE, AND ELECTRONIC APPARATUS FOR GENERATING THREE-DIMENSIONAL HUMAN FACE IMAGE FROM OCCLUDED IMAGE

摘要

A method, device, and electronic apparatus for generating a three-dimensional human face image from an occluded image. The method comprises: recognizing a target human face image in an image, and determining whether a rotation angle of a target human face is greater than a first preset threshold (S101); if the rotation angle is greater than the first preset threshold, recognizing feature point information in the target human face image by means of a pre-trained regression model, wherein first feature point information comprises first visible feature point information and first non-visible feature point information (S102); and inputting the first feature point information into a pre-trained convolutional neural network model, so as to obtain a three-dimensional human face image corresponding to the target human face image (S103). By the above method, the device, and the electronic apparatus, issues of difficult feature point recognition and location and three-dimensional human face image generation caused by self-occlusion are resolved, and three-dimensional human face image generation is further improved, improving the degree of matching.