改进匹配算法在LED粘片设备图像识别中的应用  

Application of Improved Match Algorithm in LED Die Bonder Image Recognition

作　　者：曹占伦[1] 吴小洪[1] 姜永军[1] 林晓新[1] 

机构地区：[1]广东工业大学,广州510090

出　　处：《半导体技术》2008年第3期200-203,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);广东省攻关项目(2005A1040304);高等学校博士学科点专项科研基金(20040562005)

摘　　要：在全自动LED粘片设备中,为了实时快速精确地识别出LED芯片,提出了一种新的金字塔-序贯相似性检测匹配算法,有机地结合了金字塔算法的快速性和序贯相似性检测匹配算法精确性的优点。该算法先采用金字塔算法对目标图像进行分层搜索,呈几何级的减少匹配次数,提高效率,然后用序贯相似性检测匹配算法在很小的区域内进行精匹配,达到准确定位的目的。目前,该算法已在VC++编程环境中编程实现,并在样机上得到应用,达到了工业应用的目的。A sequential similarity detection Pyramid match algorithm for real time accurate recognition of LED chip image was presented. The improved algorithm combined the rapidity of Pyramid algorithm with the accuracy of sequential similarity detection match algorithm （SSDA）. The algorithm detected the target image hierarchically with the Pyramid match algorithm, the detection times of which decreased geometrically. The Pyramid algorithm improved the efficiency of processing image. Then the SSDA was used to match in a small circumscription to confirm the position of the LED chips. The algorithm is realized in VC ＋ ＋ language and is applied currently in the image recognition system of automatic LED die bonder.

关 键 词：发光二极管 金字塔算法 序贯相似性算法 模板匹配 

分 类 号：TN391.4[电子电信—物理电子学]