Human body three-dimensional model reconstruction method, device and storage medium

Abstract

The embodiment of the invention discloses a method, device and a storage medium for reconstructing a three-dimensional model of a human body. The embodiment of the invention can acquire two-dimensional images of a human body to be reconstructed, mapping a two-dimensional image to a voxel space to obtain a voxelized result of the two-dimensional image, and then calculating the voxelized result by adopting a preset three-dimensional parameter regression network, and reconstructing a three-dimensional model of a human body based on the calculated three-dimensional deformation model parameters ofthe two-dimensional image; this scheme can map two-dimensional image to voxel space before calculating three-dimensional deformation model parameters, so it can avoid the problem of poor accuracy caused by two-dimensional image information loss, and can greatly improve the accuracy and effect of three-dimensional model reconstruction.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a method, device and storage medium for reconstructing a three-dimensional model of a human body. Background technique [0002] Three-dimensional model reconstruction is one of the important research topics in computer vision and computer graphics, and it is also one of the basic technologies of virtual reality (VR, Virtual Reality), augmented reality (AR, Augmented Reality) and mixed reality (MR). , how to better realize 3D model reconstruction is of great significance. [0003] Traditional 3D model reconstruction is mainly realized by using technologies such as depth sensors, structured light (Structure Lights), and multi-eye stereo matching (Stereo Matching). Not only is the equipment expensive, but the image collection process is also relatively cumbersome. For this reason, another solution has been proposed in the prior art, which only needs to use a single two-dimen...

AI Technical Summary

Problems solved by technology

However, in the process of research and practice of the prior art, the inventors of the present invention found that this solution is very dependent on accurate face feature point detection, and because the two-dimensional face image itself will be affected by side faces, large expressions, or Part of the information is lost due to occlusion and other situations. Therefore, in addition to the good reconstruction effect of the frontal face without occlusion, the reconstruction effect of side faces, large expressions, or faces with occlusion is not very ideal, and the accuracy is relatively low. Low

Images