Real-time 3D reconstruction method for dynamic scenes based on binary feature encoding matching

Abstract

The present invention proposes a real-time three-dimensional reconstruction method of a dynamic scene based on binary feature coding matching, wherein the method includes: using an RGBD camera to photograph the dynamic scene, obtaining a sequence of color depth images, and converting a single frame in the sequence of color depth images The image is converted into a 3D point cloud; the matching point pairs between the 3D point cloud and the vertices of the standard model are obtained through binary feature coding matching, wherein the attitude parameters of the standard model are based on the first frame of the color depth image in the color depth image sequence Determine; determine the energy function corresponding to the set corresponding to the matching point pair, and construct and update the non-rigid motion field parameters from the standard model to the 3D point cloud according to the energy function; update the standard model according to the non-rigid motion field parameters to obtain the target model. Therefore, the real-time performance, robustness and accuracy of reconstruction of dynamic scenes can be effectively improved.

Description

technical field [0001] The invention relates to the technical fields of computer vision and computer graphics, in particular to a real-time three-dimensional reconstruction method for dynamic scenes based on binary feature coding and matching. Background technique [0002] 3D reconstruction of dynamic scenes is a fundamental and important problem in the fields of computer vision and computer graphics. High-quality 3D reconstruction of dynamic scenes has broad application prospects and important application value in virtual reality, sports games, film and television entertainment and other fields. [0003] In the related art, systems based on laser scanners or multi-camera arrays require the object to be scanned to remain absolutely still during the scanning process, which makes it difficult to handle dynamic scenes, and such devices are expensive and difficult to popularize in the daily life of ordinary people. Existing reconstruction methods based on RGBD cameras are usual...

AI Technical Summary

Problems solved by technology

[0003] In related technologies, systems based on laser scanners or multi-camera arrays require the object to be scanned to remain absolutely still during the scanning process, which makes it difficult to handle dynamic scenes, and such equipment is expensive and difficult to popularize in the daily lives of ordinary people.

Existing RGBD camera-based reconstruction methods are usually based on the assumption that the object's motion range is small, and it is difficult to complete high-precision reconstruction of fast-moving non-rigid objects

Images