Motion capture method based on asynchronous video

Abstract

The invention discloses a human body motion capture method based on an asynchronous video, which can effectively recover the three-dimensional motion of a character through multiple segments of asynchronous videos of the target character. In order to utilize multiple segments of asynchronous videos, the invention provides a video synchronization and motion reconstruction method. The method comprises the steps of synchronizing multiple video segments based on a three-dimensional human body posture; and performing motion reconstruction based on the synchronized video, and modeling motion differences under different viewpoints by using low-rank constraints to realize high-precision human body motion capture from multiple segments of asynchronous videos. According to the invention, more accurate motion capture is carried out by using a plurality of non-synchronous videos. In order to utilize a plurality of non-synchronous videos, the invention provides a video synchronization method and amotion reconstruction method, and the motion difference of different videos is modeled, so that high-precision motion capture is realized.

Description

technical field [0001] The invention belongs to the technical field of computers, in particular to a motion capture method based on asynchronous video. Background technique [0002] The motion capture of the human body from asynchronous video, the existing method that can deal with this problem is only a single-view motion capture algorithm, but the inherent depth uncertainty and self-occlusion problems of a single view make this type of algorithm unable to achieve the accuracy of multi-view reconstruction . The traditional multi-view reconstruction algorithm can only solve the situation where all cameras record the same scene, but cannot handle the situation where the cameras record different scenes and the actions between videos are not completely consistent. Contents of the invention [0003] The purpose of the present invention is to address the deficiencies in the prior art, and propose a method for video synchronization and human body reconstruction based on multipl...

AI Technical Summary

Problems solved by technology

[0002] The motion capture of the human body from asynchronous video, the existing method that can deal with this problem is only a single-view motion capture algorithm, but the inherent depth uncertainty and self-occlusion problems of a single view make this type of algorithm unable to achieve the accuracy of multi-view reconstruction

The traditional multi-view reconstruction algorithm can only solve the situation where all cameras record the same scene, but cannot handle the situation where the cameras record different scenes and the actions between the videos are not completely consistent.

Images