相机圆形点阵标定法中投影点的精准定位

Accurate positioning of projection points in camera circular dot array calibration method

针对圆形特征点难以精确定位投影点的问题,提出样本点在椭圆上的最近邻点、椭圆的逼近程度和投影点集的相似性,用最小二乘法拟合椭圆投影点的最近邻点得到投影椭圆,以其中心做为投影点(简称优化最小二乘法)。拍摄有494个圆形点的点阵板,分别用灰度重心法、扫描最小二乘法、Canny边缘最小二乘法和优化最小二乘法获得不同投影点集边缘的拟合椭圆,优化最小二乘法的拟合椭圆具有最高逼近程程度,达到0.812968;用这些方法的投影点进行相机标定,测量物件长度,优化最小二乘法的测量精度最高,达到99.99%。不论是投影椭圆的逼近程度还是应用标定参数测量相同物件的测量精度,优化最小二乘法都具有最优结果,因而优化最小二乘法能够精确定位点阵标定法中的圆形投影点。

Aiming at the problem that it is difficult for circular feature points to precisely locate projection points,the nearest neighbor of sample points on ellipse,the approximate degree of ellipse and the similarity of projection point set are proposed.494 circular dots are photographed and the fitting ellipses of the edges of different projection points are obtained by gray gravity centre method,scanning least squares method,Canny edge least squares method and optimized least squares method respectively.The fitting ellipse of the optimized least squares method has the highest approximation range which reaches 0.812968.Camera calibration is carried out with the projection points of these methods to measure the length of the object and optimize the ellipse.The measurement accuracy of the optimized least square method is the highest which is 99.99%.Whether the approximation degree of projection ellipse or the measurement accuracy of the same object by using calibration parameters,the optimized least square method has the best result.So the optimized least square method can accurately locate the circular projection points in the lattice calibration method.