摘要
大口径光学组件面形检测系统中,激光光斑的分割、质心提取和背表面光斑的自动剔除算法关系到系统的检测精度。针对系统中多路光斑光强差异比较大的现象,提出了两次分割法,有效定位各光斑区域;通过比较不同质心提取算法的稳定性,选取灰度重心法提取激光光斑质心;采取基于特征的分类技术识别元件前表面反射光斑。实验结果表明,算法能精确稳定地提取光斑质心和有效识别前表面光斑,该方法已商用于光学组件面形检测系统。
In the large optical components topography measurement system, the accuracy of measurement is greatly related to laser spot segmentation, centroid extraction and back surface spot removal.As the great difference in multi-spots＇ light intensity, a twice-segmentation method is proposed to locate the region of the laser spots efficiently.Compared the stability of the algorithms for laser spot＇s extraction, a gray-gravity method is chosen.A feature based classification method is proposed to discriminate the spots which are reflected by front surface of optical component.Experimental results demonstrate that these algorithms can extract laser spots accurately, stably and classify laser spots efficiently.Also, they are used in optical components topography measurement system successfully.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第20期163-166,共4页
Computer Engineering and Applications
基金
国家自然科学基金委-中国工程物理研究院联合基金项目(No.10676029
No.10776028)
关键词
元件面形
激光光斑
两次分割法
质心提取
前表面光斑识别
optical components topography measurement
laser spot
twice-segmentation
centroid extraction
discrimination of front reflected laser spot
作者简介
张隽楠（1985-），男，硕士研究生，主要研究领域为机器视觉、图像处理；E-mail：cplusplus2010@163．com
范勇（1972-），男，博士，副教授；
陈念年（1977-），男，讲师。
