Three-dimensional reconstruction method of scene spots in binocular stereo visual system

Abstract

A three-dimensional reconstruction method of scene spots in a binocular stereo visual system comprises a binocular camera formed by two cameras. Firstly, the binocular camera is calibrated, and an optical center OA and an optical center OB of the binocular camera are calculated; then the scene spots are matched with image points obtained from an imaging plane of the camera, and the coordinates PA and PB of a corresponding point in the imaging plane are obtained; two radials formed by the optical centers of the binocular camera and the corresponding image points are intersected at a point P which is a primary image point of the image points, the optical centers of the binocular camera and the primary image point of the image points serve as control vertexes, and a quadratic Bezier curve is generated; a degree elevation treatment is carried out on the quadratic Bezier curve to obtain a cubic Bezier curve, and the optical centers of the binocular camera and the primary image point of the image points serve as control vertexes of the cubic Bezier curve. Due to the fact that the quadratic Bezier curve and the cubic Bezier curve treated in a degree elevation mode represent the same curve in the space, the quadratic Bezier curve and the cubic Bezier curve are combined into an equation set, and then the coordinates P of the primary image point in the space are solved. The three-dimensional reconstruction method of the scene spots in the binocular stereo visual system can directly, simply and accurately determine two three-dimensional coordinate primary images of the image points.

Description

technical field [0001] The invention relates to a depth measurement and three-dimensional reconstruction method of a field spot in a stereo vision system, and is suitable for a three-dimensional reconstruction method of a field spot in a binocular stereo vision system in computer vision and photogrammetry. Background technique [0002] The binocular stereo vision system mainly calibrates the two cameras first to obtain the internal and external parameters of the cameras; then through the corresponding image point coordinates of a scene point in the two cameras, and by fusing the scene points observed by the two cameras , calculate the three-dimensional coordinates of the scene points in space, and reconstruct these scene points to obtain the three-dimensional original image. [0003] When performing 3D reconstruction, the original image of the corresponding point on the two images theoretically appears at the intersection of the ray passing through the imaging point of the c...

AI Technical Summary

Problems solved by technology

This method is more complicated in the process of finding the line segment intersecting with the ray and perpendicular to it, and often does not get the optimal solution.

Images