Video processing method, device and electronic equipment

Abstract

Embodiments of the present invention provide a video processing method, device, and electronic equipment, the method comprising: in response to a user's instruction, acquiring the first human face image in the target picture and the second human face image to be replaced in the video; using The convolutional neural network model outputs the first three-dimensional human face image corresponding to the first human face image; obtains the expression parameters of the second human face image, and adjusts the first three-dimensional human face image according to the expression parameters , to obtain a second three-dimensional face image; mapping the second three-dimensional face image to a two-dimensional space to obtain a target face image, and replacing the second face image with the target face image. Through the above method, device and electronic equipment, there is no need to collect multiple pictures from multiple angles, and only a single picture is needed to replace the face in the video, so as to obtain the target face; at the same time, considering the influence of facial expression factors, it is possible to improve The correlation between the target face and the video.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a video processing method, device and electronic equipment. Background technique [0002] In recent years, the generation technology of virtual objects has been more and more applied in video production, among which, the reconstruction of 3D face is particularly important for the generation of virtual objects. In the process of implementing the present invention, the inventor found that in the prior art, the human face is generally used to replace the target human face in the video to realize the generation of the virtual object, but the expression factor of the face is not considered so that the association between the virtual object and the video The performance is poor, which leads to the fact that the generated virtual objects do not match the video environment, and the user feels false. [0003] In addition, the generation of virtual objects in video produc...

AI Technical Summary

Problems solved by technology

The two-dimensional face image is mainly obtained through the camera of the mobile phone, and the effect of the reconstructed three-dimensional face image partly depends on the image quality obtained by the previous camera device, and the obtained image quality partly depends on the effect of the mobile phone on the shake processing when shooting , the current mobile phone mainly performs anti-shake processing through software, and there are not many measures for hardware improvement

Images