Aiming at the difficulty and low precision of the turnover location in quick PCB system carved on basis of physics, the method of vision alignment in subpixel is researched. Using Hough Transform can get the coordinate of PCB mark in pixel, and filtering the disturbance point. Then circle fitting use least-squares method to getting subpixel coordinate. According to deviation of the actual location and the theoretical position corrected the location data with the homogeneous transfer matrix. Experiments show that vision alignment has solved the problem of the low precision and complex operation.%针对基于物理雕刻的快速PCB制作方法中存在的翻板定位精度不高的问题,进行了PCB定位孔亚像素视觉定位算法的研究.通过Hough变换得到定位孔的边缘像素点并过滤掉非圆上的干扰点,再使用最小二乘法进行圆拟合实现定位孔的亚像素定位以提高系统的定位精度.根据实际位置与理论位置偏差值得出校正加工数据的齐次变换矩阵,实现翻板定位时的数据校正.实验证明,视觉定位很好的解决了翻板定位精度不高,操作复杂的问题.
展开▼