A GPU-based and depth-oriented real-time 3D reconstruction method

Abstract

The invention discloses a GPU-based real-time three-dimensional reconstruction method facing a depth image. The method executed on a GPU comprises a data preprocessing stage, a data registration stage, a data fusion stage and a three-dimensional model extraction stage. At the data preprocessing stage, data truncation, data filtering and data layering are carried out successively; at the data registration stage, a pose of depth image data in a global coordinate is solved and obtained; at the data fusion stage, the depth image data are fused into an improved three-dimensional truncation directed distance field model; and at the three-dimensional model extraction stage, a three-dimensional grid model is extracted from the n improved three-dimensional truncation directed distance field model. Therefore, with effective utilization of the characteristic of GPU multi-thread parallel processing, real-time three-dimensional reconstruction is realized on the GPU, so that real-time three-dimensional reconstruction is realized on the premise that high precision is guaranteed.

Description

technical field [0001] The invention relates to the field of computer graphics, in particular to a GPU-based and depth image-oriented real-time three-dimensional reconstruction method. Background technique [0002] Real-time 3D reconstruction technology is used to obtain 3D models of objects in real space in real time. 3D reconstruction technology is widely used in multiple directions involving computer technology, including computer animation, computer-aided design, robotics, virtual reality, augmented reality, 3D print etc. After years of development, 3D reconstruction technology has made great progress, but there are still some unresolved problems in 3D reconstruction, such as reconstruction speed, reconstruction range, and reconstruction accuracy. [0003] The main difficulty of an effective real-time 3D reconstruction method is to achieve real-time processing effect while maintaining the reconstruction accuracy, and at the same time, the reconstruction range must be la...

AI Technical Summary

Problems solved by technology

Although the emergence of depth cameras can obtain high-precision depth data, there are still many problems in real-time 3D technology. With the development of 3D technology, a large number of calculations involved in 3D technology began to be operated on the GPU, so that the real-time 3D reconstruction technology saw a glimmer of hope, but there are still problems, such as the 3D reconstruction has to deal with a large amount of data, and the requirements for video memory are very high. Large, so there are still problems in dealing with very large scenes

Images