A texture fusion method for real-time 3D reconstruction of RGB-D camera

Abstract

The invention discloses a texture fusion method for real-time three-dimensional reconstruction of an RGB-D camera. Process the RGB-D data stream to obtain the clarity of the color image, select the key frame to extract the foreground, filter and denoise the depth image, calculate the normal vector of the point cloud of the depth image, and construct the reconstruction data stream; use the probabilistic method and the heuristic method The combined method quantitatively establishes the adaptive weight field of the color image as the confidence distribution of the real-time frame describing the color data; by comparing the confidence weight in the adaptive weight field of the real-time frame with the latest confidence weight of the reference point cloud, Choose from the three operations of replacement, fusion and preservation to update the texture result and realize texture fusion for 3D reconstruction. The invention can extract high-quality data, can effectively reduce the blurring of texture fusion, realize clear texture reconstruction results, and embed it into the RGB-D reconstruction framework at a low calculation cost, thereby significantly improving the texture reconstruction accuracy.

Description

technical field [0001] The invention belongs to the field of computer vision and computer graphics image processing, and more specifically relates to a texture fusion method for real-time three-dimensional reconstruction of an RGB-D camera. Background technique [0002] With the increasing importance of 3D reconstruction technology in the fields of autonomous driving, virtual reality, robot positioning and navigation, how to use low-cost RGB-D sensors to obtain high-quality 3D reconstruction scenes in real time has become a hot issue. However, most of the current real-time 3D reconstruction based on RGB-D adopts the method of pixel-by-pixel weighted average sequence frame images to generate the texture of the object surface. This method is easy to destroy the structural information of the texture and cause the texture blur problem. [0003] At present, there are two main ways to improve the quality of texture reconstruction. The first way is to select high-quality data as m...

AI Technical Summary

Problems solved by technology

This method usually requires good initial reconstruction results, and at the same time requires high computing costs, it is difficult to realize real-time application on ordinary PCs or mobile terminals, and it is not easy to integrate into the existing three-dimensional reconstruction framework

Images