提高大型圆截面形状测量精度的方法

Method for Improving Measurement Precision of Large-Scale Circular Section

由于难以均布采集坐标点,导致常规最小二乘法拟合的精度低,不适于大型工件.实验结果表明半径和圆心拟合结果之间存在线性依赖关系.因此分别从提高圆心定位精度和半径测量精度两方面提高测量精度.用双目视觉传感器组成配对网络,利用平行弦方法提高圆心定位精度.基于设计半径已知条件,利用半径约束最小二乘法提高测量精度.对大型钢管工件和隧道构件等圆形截面对象进行仿真和实验,采用平行弦方法将圆心偏差由0.005mm降至0.003mm,采用半径约束方法将圆心偏差由25.24mm降至5.06mm.结果表明,两种方法均可有效提高圆拟合精度,对噪声具有较好的鲁棒性.

The data points are distributed asymmetrically, the ordinary least square method is unfit for large-scale work-pieces due to its low precision. According to the theoretical analysis and practical measurement, there is a linear relationship between the fitting results of the circle center and the radius. In this paper, a new method for improving measurement precision based on the above conclusion is presented. The matching network composed of binocular stereo vision sensor is introduced, and the orientation precision of the circle centre is advanced by parallel chord method. Besides, the radius precision is ac- complished by least square method with radius constraint. Measurements of steel work-pieces and tunnel components are carried out by using simulation and experiment, and the center orientation error decreases from 0. 005 mm to 0. 003 mm by means of parallel chord method, and from 25.24 mm to 5.06 mm by use of least square method with radius constraint, which indicates that two proposed methods can improve measurement precision of circle effectively and have better robust to noise.