大口径平面光学元件面形检测中激光光斑质心定位

如何快速准确计算透镜焦面上的光斑质心坐标是大口径平面光学元件面形检测系统的核心问题之一，直接决定面形检测的精度和重复性。定义一个半径为r的探测窗口，根据到中心距离建立一维卷积模板，在光斑覆盖区域中寻找窗口内能量和局部最大的位置，确定光斑有效区域；再使用传统重心法计算有效区域质心；最后应用3准则进行误差处理。实验结果表明，该算法检测精度为0.1像素，各指标比传统阈值加权质心定位法提高约1 倍，面形测量的面形相似性和PV 比阈值加权法更贴近干涉仪检测结果。目前，通过长时间的实验验证与改进，已成功运用于实际项目中。

Centroid detection of laser spots in large plane optical components topography measurement system

One of the core issues of the large plane optical components topography measurement system is calculating the centroid coordinates of laser spots quickly and accurately, which determines the accuracy and repeatability of the measurement. According to visual perception, a 1D detection window with radius as r was defined, to find the local maximum energy in the field with a convolution template depending to center distance. Then traditional centroid method was used to calculate the subpixel centroid. Finally, 3 rule of error processing was applied. The experimental results show that the detection accuracy of proposed algorithm is 0.1 pixel, and the index was improved about 1 time than existing threshold weighted centroid method. The similarity and PV of topography is closer to interferometer than threshold weighted measurment. It has been successfully applied in actual projects.