Multi-view three-dimensional reconstruction method based on catadioptric panoramic camera

Abstract

The invention discloses a multi-view three-dimensional reconstruction method based on a catadioptric panoramic camera, belongs to the technical field of three-dimensional reconstruction, and aims to solve the problem of low three-dimensional reconstruction integrity of a weak texture region. The method comprises the following steps: acquiring a sequence image by using a catadioptric panoramic camera, and masking redundant information in the sequence image; solving an internal reference; calibrating the panorama camera by using matlab so as to obtain the internal reference of the catadioptric panorama camera; solving a rotation matrix R and a translation vector t between the cameras; dense point cloud reconstruction and optimization, wherein dense point cloud reconstruction is carried out through a multi-view stereoscopic vision algorithm; pixel points of the image are matched, and dense point clouds capable of meeting reconstruction requirements are obtained; carrying out point cloud depth information fusion; solving depth information of the target object by using a photometric stereo algorithm, and fusing the depth information with the solved point cloud; the problem of low integrity of three-dimensional reconstruction of a weak texture region is improved; mesh mapping is carried out by using meshlab; and displaying a three-dimensional effect.

Description

technical field [0001] The invention relates to a multi-view three-dimensional reconstruction method based on a catadioptric panoramic camera, belonging to the technical field of three-dimensional reconstruction. Background technique [0002] In the 3D reconstruction of ordinary cameras, when the texture is too small or unclear, the number of feature points is not enough, and the obtained spatial transformation matrix is ​​inaccurate or even impossible to obtain, which may seriously lead to the failure of the entire reconstruction task. Light ones will also lead to larger errors in the reconstruction results. The catadioptric panoramic camera ensures sufficient feature points, and there is also a certain overlapping area between adjacent panoramic images. It is easy to establish a spatial relationship between the images, and the shooting position and angle do not need to be considered when shooting. A large area of ​​reconstruction can be completed by realizing a small numb...

AI Technical Summary

Problems solved by technology

[0004] In order to effectively solve the problem of low integrity of 3D reconstruction in weak texture areas, the present invention proposes a multi-view 3D reconstruction method based on a catadioptric panoramic camera

Images