新图像匹配算法在金丝球引线键合机中的应用  

Application of Novel Image Matching Algorithm in Golden Wire Bonder

作　　者：吴小洪[1] 回晓明[1] 姜永军[1] 钟石明[1] 

机构地区：[1]广东工业大学机械装备制造及控制技术教育部重点实验室,广州510090

出　　处：《半导体技术》2009年第7期631-633,640,共4页Semiconductor Technology

基　　金：国家自然科学基金(50475044);教育部科技研究重点项目(2004106);广东省科技计划攻关项目(2006A10401003);广州市科技攻关项目(2006Z3-D9071);高等学校博士学科点专项科研基金(20040562005)

摘　　要：随着当今社会对金丝球引线键合机的工作速度、定位精度要求的不断提高,对图像控制系统也提出了更高的要求。针对单纯使用MIL图像处理库提供的模板匹配算法会出现误判现象,提出了一种"二次梯形分层搜索的序贯相关判决算法"作为MIL匹配算法的补充,即对一幅图像实施一主一辅进行两次模板匹配运算的方法,使图像的匹配既能实现快速准确又不出现芯片误判的情况。目前该算法已在VC++中编程实现,并在工业上得到应用,结果表明此算法是高效可靠的。With the increasing demands on the accuracy of positioning and working speed of the gold wire bonder, it set a higher demand on the PR system. Aimed at the phenomenon of misjudgment by simply using the pattern matching algorithm in MIL image processing library, a kind of algorithm was adopted called ＂sencondry sequential similarity detection algorithm of hierarchical search＂ as a supplement of MIL pattern recognition algorithm. It can raise the accuracy of positioning, working speed and avoid the misjudgment by using two times template matching through taking a primary and a subsidiary image to matching a picture. At present, the algorithm is fulfilled in VC ＋ ＋ and used in the industry. The result proves that the algorithm is efficient and reliable.

关 键 词：金丝球引线键合机 序贯相似性检测算法 图像匹配算法 芯片 误判现象 

分 类 号：TN391.4[电子电信—物理电子学]