双目视觉测量系统特征点提取与匹配技术研究

自主研发了一种光笔式双目立体视觉大工件尺寸测量系统，对测量系统中特征点的提取及其匹配技术进行了研究。测量系统采用 Canny 算子和 Zernike 矩相结合的算法实现椭圆光斑的亚像素边缘提取，根据得到的亚像素边缘点，采用基于最小二乘的曲线拟合法得到椭圆光斑的中心坐标。针对特征点的匹配问题，提出了一种基于位置约束的快速匹配方法。实验结果显示:所提方法能提取到椭圆光斑亚像素边缘，可精确计算出椭圆光斑中心坐标，匹配率达到95%以上。

Feature points extraction and matching technology of binocular vision measurement system

A light pen based binocular stereo vision measurement system for large scale workpiece was developed,in which feature points extraction and matching technology were studied deeply. The sub pixel edge of ellipse spot was achieved accurately by the algorithm based on Canny operator and Zernike moments. Based on the ellipse spot edge points, the curve fitting method based on the least square was used to realize the center extraction of the ellipse spot. Aiming at the matching problem of feature points,the matching method based on position constraint was proposed.Experiment results demonstrate that the proposed method can extract the sub pixel edge of the ellipse spot, accurately calculate the center coordinate of the ellipse spot, and the matching rate is greater than 95%.