基于共线性的机械臂视觉系统畸变矫正

Distortion Correction of Manipulator Vision System Based on Collinearity

对机械臂视觉系统在进行动态测量分析时,通常因镜头畸变产生定位误差,对此问题进行了研究。提出了一种基于共线性的机械臂视觉系统畸变矫正方法。该方法在共线性原则的基础上,根据直线投影不变性原则和径向畸变模型的关系,对同一直线上的特征点坐标构造目标函数;利用Levenberg-Marquardt(LM)算法,对目标函数进行迭代寻优,获得全局畸变系数最优解。通过与传统方法比较发现,文章提出的方法得到的k1、k2的相对误差分别为-4.8%和-5.1%。与传统方法相比具有较好的可靠性。

A location error will be generated by the lens distortion when the vision system of a manipulator works.A lens distortion correction method based on collinearity is proposed to calibrate the parameters of the vision system.First,an objective function of points on a line is built based on the principle of constant linear projection and the radial distortion relationship.Second,the objective function is optimized by Levenberg-Marquardt(LM)algorithm to get the optimum estimate of the lens distortion coefficients.A lens distortion correction method is proposed to calibrate the parameters of the vision system.The relative errors between the calculated distortion coefficient k1、k2 and traditional methods are-4.8% and -5.1% respectively.The proposed correction method has a better reliable compared to the traditional methods.