基于NCC改进算法的SMT电路器件定位  

Circuit element location in SMT based on improved NCC algorithm

作　　者：邢培锐 童敏明[1] 张俊升[2] 孙苏园 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221008 [2]中国人民解放军第五七一五工厂,河南洛阳471000

出　　处：《计算机应用》2017年第A01期194-197,共4页journal of Computer Applications

基　　金：国家重点研发计划项目(2016YFC0801808)

摘　　要：随着电子产品的微型化发展,对表面贴装技术(SMT)中电路器件的定位精度和速度提出了越来越高的要求。针对SMT贴合工艺中标记点匹配速度慢、边缘检测精度差的问题,为了提高模板匹配效率,在经典NCC算法的基础上提出了一种缩小搜索区域的NCC算法(RSR-NCC),通过提取模板图像中关键信息与搜索图像进行匹配,实现电路器件的定位。在保证匹配准确性的情况下,该算法可减少计算量,同时增加关键信息的权重。分别采用传统NCC算法、SSDA算法、NCC+粗搜索算法与RSR-NCC算法对SMT贴合过程中随机采集的图像进行匹配定位,结果显示RSR-NCC算法在保证匹配结果准确的基础上,匹配时间分别为NCC算法、SSDA算法、NCC+粗搜索算法的1%、2.8%、8.5%,明显提高了匹配速度,证明了该算法的优越性。With the miniaturization of electronic products development, higher demands are proposed in accuracy and speed of the circuit element locating in the Surface Mounted Technology（ SMT）. There are many disadvantages in the SMT at present, which including slowly marked points matching and poor precision of edge detection. In order to improve the efficiency of template matching, a new algorithm called Reduce Search Region-Normalized Cross Correlation（ RSR-NCC） was proposed based on traditional algorithm of NCC, by extracting the key information of the template image and comparing it with the searched image to locate the circuit, while guaranteeing the accuracy, the amount of calculation reduced and the weight of the key information increased. The randomly captured images in SMT autolaminating machine were used to perform the matching experiment. The matching time of improved algorithm was 1%, 2. 8%, 8. 5% of NCC, SSDA and NCC-coarse search algorithms. RSR-NCC brings an obvious improvement of matching efficiency.

关 键 词：SMT NCC 标记点定位 图像匹配 粗搜索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]