Three-dimensional face reconstruction method, electronic equipment and computer readable storage medium

Abstract

The embodiment of the invention provides a three-dimensional face reconstruction method, electronic equipment and a computer readable storage medium, and relates to the technical field of image processing. According to the embodiment of the invention, the target face in the two-dimensional face image is divided into the first part which does not need to be corrected and the second part to be corrected, and after the three-dimensional point cloud of the target face is obtained through prediction, the three-dimensional point cloud of the second part in the target face is corrected to obtain the three-dimensional point cloud of the corrected target face. Thus, the method can correct the second part with the large motion range under the condition that the overall face posture predicted in the three-dimensional face prediction network model fits the actual face shape, optimizes the fitting effect of the prediction result of the part with the large motion range and the actual part, and the fitting effect of the predicted human face overall posture and the local part is considered.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular, to a three-dimensional face reconstruction method, electronic equipment, and a computer-readable storage medium. Background technique [0002] In live broadcast application scenarios, there will be some requirements for 3D animation effects, such as rabbit ears, pig masks, and clown hats. These 3D animation effects need to be based on 3D face reconstruction. 3D face reconstruction includes not only The location of some key points of the face also includes the state information of the face at this time, such as the rotation angle information of the face. In addition, 3D face reconstruction may have some personalized requirements in some scenarios. For example, after the user takes a self-portrait, a 3D cartoon image with a high similarity to the user is generated according to the self-portrait of the user. [0003] Existing 3D face reconstruction schemes inc...

AI Technical Summary

Problems solved by technology

[0003] Existing 3D face reconstruction schemes include schemes based on deep learning and schemes based on regression coefficients. Among them, the scheme based on deep learning can make the predicted overall pose of the face fit better with the actual face shape, but it has a large range of motion. parts (for example, mouth or eyes), the prediction results are poorly fitted with the actual parts

The traditional regression coefficient scheme can better achieve the fitting effect between the predicted result of the part with a large range of motion and the actual part, but the fitting effect between the predicted overall pose of the face and the actual face shape is poor.

Images