摘要
在双目立体视觉技术中,经过特征提取和立体匹配得到两两对应的像素点对,根据摄像机标定建立的物体空间坐标到像平面坐标对应的矩阵,利用最小二乘法求取物体三维空间点的坐标。但是,由于最小二乘法没有考虑所建立的超限定方程组所代表的几何意义,计算出的点坐标精度不高。本文提出一种考虑方程组所代表几何意义的方法,利用异面直线公垂线中点去逼近物体空间点。
In binocular stereo vision, after character distilling and three-dimensional matching, there are many two pixel points which be matched with each other, according to camera calibration to create a matrix and use least square-method to obtain coordinate point which is three-dimensional. But least square-method do not consider geometric meaning that the equation is represent for, so the measure precision is lower. A model is described which consider the equation geometric meaning, making use of mid-point of vertical line in crossover lines to get three-dimensional points coordinate.
出处
《机电产品开发与创新》
2008年第5期31-32,35,共3页
Development & Innovation of Machinery & Electrical Products
关键词
双目立体视觉
摄像机标定
立体匹配
最小二乘法
异面直线公垂线中点
binocular stereo vision
camera calibration
three-dimensional matching
least square-method
mid-point of vertical line in crossover lines
