一种点目标三维重构的新算法

双目立体视觉研究利用二维投影图像恢复三维景物世界的问题,其中三维重构是立体视觉测量过程的主要组成部分之一,是从二维图像恢复三维景物的关键环节.针对空间点目标的三维重构,传统的方法是采用最小二乘法进行的,但是由于最小二乘法没有考虑所建立的超定方程组所代表的几何意义,所以计算出的点坐标精度不高.对此,从几何学的角度出发,提出了一种异面直线公垂线段中心法来提高点目标的重构精度.实验表明,该方法比传统的最小二乘法具有更高的测量精度.

New Algorithm for Three-dimensional Construction of Point Object

The problem that two-dimensional projection images are used to recover three-dimensional world is studied mainly in binocular stereo vision,in which three-dimensional construction is one of the important parts,and it is the key from two-dimensional image to three-dimensional environment.Aiming at three-dimensional construction of point object,lease square-method is usually used in traditional method.But because it don't consider geometric meaning that the over-determined equation set is represent for,the precision of point object measured is low.To this,from the view of geometrical mathematics,a kind of method based on middle point of crossover line common vertical line is proposed to improve reconstruction precision of point object.Compared with lease square-method,the test indicates that the method has the higher measurement precision.