A high-precision 3D reconstruction method and system for depth cameras

Abstract

The invention discloses a depth camera-oriented multi-view three-dimensional reconstruction method and system. The method includes: acquiring a plurality of pairs of images of a real scene from different angles according to a depth camera, each pair of images including a color image and a depth image; acquiring two-dimensional feature points and feature descriptors on each color image according to a feature point detection algorithm, and calculating The confidence between the feature descriptors corresponding to any two color images is obtained to obtain the feature matching relationship; according to each feature matching relationship and the corresponding depth image, the corresponding 3D image is obtained, and the 2D feature points are back-projected to the camera coordinate system Next, obtain the 3D points on the 3D image; calculate the absolute pose of the camera according to the 2D feature points and the 3D points; according to the absolute pose of the camera, map the 3D points to the camera coordinate system to obtain the initial 3D model of the scene. By adopting the method or system of the present invention, a high-precision three-dimensional scene model can be restored from the image sequence collected by the depth camera.

Description

technical field [0001] The invention relates to the fields of computer vision and computer graphics, in particular to a high-precision three-dimensional reconstruction method and system for depth cameras. Background technique [0002] High-precision 3D reconstruction has always been one of the hot research issues in the field of computer vision and graphics, and its purpose is how to efficiently restore the 3D model of the scene from the input image sequence. At the same time, 3D reconstruction technology plays an irreplaceable role in the game industry, military simulation, agriculture and industry, augmented reality, virtual reality, scene monitoring, map navigation and automatic driving. Due to the influence of factors such as illumination changes in the scene, occlusion between objects, scene aggregation, repetitive structure, and noise in the collected data, it is difficult for the traditional manual-based active 3D modeling method to reconstruct the real 3D scene mode...

AI Technical Summary

Problems solved by technology

Due to the influence of factors such as illumination changes in the scene, occlusion between objects, scene aggregation, repetitive structure, and noise in the collected data, it is difficult for the traditional manual-based active 3D modeling method to reconstruct the real 3D scene model, cannot recover high-precision 3D scene model contained in a large number of image sequences

Images