基于粗糙集的印刷锡膏自动识别和定位  

Automatic Recognition and Location of Printed Solder Paste Base on Rough Set

作　　者：吴晖辉[1] 张宪民[1] 邝泳聪[1] 洪始良[1] 

机构地区：[1]华南理工大学机械与汽车工程学院,广东广州510640

出　　处：《测试技术学报》2009年第5期423-428,共6页Journal of Test and Measurement Technology

基　　金：国家杰出青年科学基金资助项目(50825504);广东省科技攻关重点资助项目(2008A010300002);粤港关键领域重点突破招标资助项目(20081628);广东省自然科学基金资助项目(8452800001000023)

摘　　要：为了准确获取贴装后印刷电路板(PCB)上锡膏的印刷位置,提出了一种综合了锡膏颜色和纹理特征的粗糙集方法.以红、绿、蓝三种颜色分量直方图特征为等价关系求得锡膏印刷区域的上近似集;以均值、方差和一致性等纹理特征为等价关系获得锡膏的下近似集;分别计算上、下近似集的相对权重,获得分割后的锡膏印刷区域的重心.实验结果表明,综合了颜色和纹理信息的粗糙集方法,定位精度为1个像素,定位速度约为10 ms,精度高,速度快.To locate the printed solder paste in the PCB （printed circuit board） after mounting, a novel rough approximation method which synthesized the color and texture features of the solder paste was proposed in this paper. First, the upper approximation of printed solder pastes was extracted with RGB （Red, Green, Blue） color attributes. Second, by extracting the texture features such as mean, variance, unity and so on from the gray image, the lower approximation of printed solder pastes was constructed. Finally, the center of gravity of the printed solder pastes was represented by computing the relative importance of the lower and upper approximations. Experimental results demonstrate that the location precision is 1 pixel and the consumed time is about 10 ms, which has improved the proposed rough set approach that synthesizes the color and texture features is accurate and time saving.

关 键 词：SMT 锡膏 粗糙集 颜色直方图 纹理特征 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]