面向大尺寸检测CCD图像中心提取精度的研究  

Study on the center extraction precision of image photographed by CCD for large scale inspection

作　　者：林润芝[1] 杨学友[1] 邹剑[1] 邾继贵[1] 吴斌[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《传感器与微系统》2010年第12期51-53,共3页Transducer and Microsystem Technologies

基　　金：天津市科技支撑计划资助项目(08ZCGYGX11800);国家自然科学基金重点资助项目(50735003);国家自然科学基金资助项目(50705065)

摘　　要：大尺寸测量中,数字图像处理技术起到了至关重要的作用,而图像中心特征的提取精度是影响数字图像处理的核心因素之一。目前常用的中心特征的提取方法是传统质心法、灰度阈值质心法、平方加权质心法和高斯曲面拟合法质心法。由于采集图像时物体距CCD像面距离不固定,上述方法对不同物距采集的图像中心特征提取效果不同。现将在理论分析和仿真实验验证2个方面对这4种方法的特性进行比较。实验结果表明:高斯拟合质心法更适合较远距离采集图像的标定,而另3种质心法更适合较近距离采集图像的标定。In large scale measurement, digital image processing technique plays an essential role. And the precision of center extraction is one of the core factors that impact digital image processing. The common methods are traditional eentroid method, gray threshold centroid method, square weighting eentroid method and gauss curve fitting centroid method. For that the results of these extraction methods at different distances are different, the characters of these four methods are compared both through theoretical analysis and simulation experiment. The results show that Gauss curve fitting eentroid method is more suitable for pictures taken from longer distance ,while the other three methods are more suitable for photos taken from shorter distance.

关 键 词：中心提取 质心法 灰度阈值 加权 高斯拟合 

分 类 号：TB852.1[一般工业技术—摄影技术]