Camera attitude optimization method for RGBD three-dimensional reconstruction texture generation

Abstract

The invention belongs to the technical field of computer vision and computer graphics image processing, and in particular relates to a camera attitude optimization method for RGBD three-dimensional reconstruction texture generation. The method comprises the steps that S1 a depth image fusion keyframe is acquired; S2 a three-dimensional mesh model point cloud depth projection map is calculated; S3background data in a point cloud depth projection map are removed; and S4 depth data matching and camera pose optimization are carried out. The method is based on the known three-dimensional mesh model and keyframe, is combined with the requirements of a texture generation process, uses the geometric features of the mesh model and depth data in the keyframe to study an optimization method for thecamera pose under the view point of the keyframe in the RGBD three-dimensional reconstruction texture generation process, and provides a guarantee for high-quality texture image generation.

Description

technical field [0001] The invention belongs to the technical field of computer vision and computer graphics image processing, and in particular relates to a camera pose optimization method for RGBD three-dimensional reconstruction texture generation. Background technique [0002] In recent years, with the continuous exploration of the potential of VR / AR applications, there has been a demand for mass production of 3D content. The 3D reconstruction technology of the scene based on the RGBD data stream enables the model to have high geometric accuracy. However, realistic 3D models also require high-quality textures to reveal details that geometric models cannot. In the process of texture generation, the accuracy of the camera pose plays a vital role in the continuity and clarity of the final 3D model texture. [0003] Most of the current research on 3D reconstruction based on RGBD data streams focuses on how to generate high-precision geometric models through RGBD data strea...

AI Technical Summary

Problems solved by technology

[0004] The above process can obtain a high-quality texture map under ideal conditions, but in actual situations, when capturing key frames, the RGB and depth data of a single frame often have holes and blur noise interference, as well as error accumulation during the modeling process It causes the camera pose to shift, which causes the RGB information to be projected onto the mesh patch, causing dislocation, causing the texture on the model to be disordered, causing irreversible effects on the subsequent seam processing, thus affecting the quality of the final texture of the mesh model

Images