面向贴片机视觉系统的摄像机标定方法  

Chip mounter vision system oriented camera calibration method

作　　者：岑誉 高健[1] 曾友[1] 简川霞[1] 

机构地区：[1]广东工业大学机械装备制造与控制技术教育部重点实验室,广州510006

出　　处：《现代制造工程》2016年第3期125-129,共5页Modern Manufacturing Engineering

基　　金：广东省高等学校科技创新项目(2012CXZD0020);广东省战略性新兴产业专项资金项目(2012A080303004)

摘　　要：针对贴片机视觉系统摄像机标定便捷性和高精度的要求,分析摄像机模型及其局限性,提出采用一阶单系数除式径向畸变模型代替传统的多项式畸变模型,并设计出带矩形边框和三角标志的圆形标定点平面标定板进行摄像机标定。通过可递归执行的修正的Deriche滤波快速获得像素级边缘,再进行亚像素边缘提取、线性方法拟合椭圆,对标定点进行精确定位,通过空间点及其图像投影的对应关系,获得较理想的摄像机参数初始值。最后,利用最优化算法进一步求解,获得更高精度的摄像机参数。实验结果表明,该标定方法简单、快速,并且精确,能满足贴片机视觉系统的要求。Under the requirement for convenience and high precision of chip mounter vision system camera calibration, the camera projection model its boundedness are analyzed. In order to make the camera calibration, a division distortion model instead of tra- ditional polynomial distortion model is proposed and a circular calibration points planar calibration board with rectangular frame and triangular mark is designed. After obtaining the pixel edge by the recursive corrected Deriche filter,the calibration points can be located precisely by extracting the sub-pixel edge and fitting ellipse with linear method, and thus the correspondence between the calibration marks and their projections in the image is determined and the reasonable initial values of camera parameters are calculated. Finally ,the higher precision camera parameters are obtained by optimization algorithm. The experimental result indi- cates that the calibration method is simple, fast and precisely, which meets the chip mounter vision system requirement.

关 键 词：贴片机 视觉系统 摄像机标定 亚像素边缘 椭圆拟合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]