QFP芯片贴装定位的视觉检测算法研究  

Research on Visual Detection APlgorithm of QFP Chip Placement

作　　者：姜利[1] JIANG Li(Changzhou Technician College Jiangsu Province,Changzhou 213032,China)

机构地区：[1]江苏省常州技师学院,江苏常州213032

出　　处：《机械制造与自动化》2022年第5期133-135,共3页Machine Building & Automation

摘　　要：针对采用数学形态学闭运算进行QFP芯片引脚修复的图像检测算法实时性差问题,设计一种基于极点四边形的QFP芯片位置误差视觉检测算法。利用QFP芯片图像的4个极值点确定极点四边形;依据极点四边形的转角初步确定芯片的转角;利用转角和4个极值点确定的4条直线提取芯片每侧引脚顶点集;采用推广的最小二乘法对芯片每侧引脚顶点集进行矩形拟合,由拟合后的直线斜率和直线交点计算出芯片转角和中心坐标。研究结果表明:该算法的运行时间约为数十毫秒,检测精度达到亚像素级,具有很好的实用性。Aimed at the poor real-time performance of image detection algorithm for QFP chip pin repair using mathematical morphology closed operation, a visual detection algorithm for QFP chip position error based on pole quadrilateral is designed. The four extreme points of QFP chip image are used to determine the polar quadrilateral, and the chip rotation angle is preliminarily determined according to the rotation angle of the polar quadrilateral. The pin vertex set on each side of the chip is extracted by the four lines determined by the rotation angle and the four extreme points, and the rectangular fitting of the pin vertex set on each side of the chip is carried out by generalized least square method. The chip rotation angle and center coordinates are calculated through fitting line slope and line intersection. The results show that the running time of the algorithm is about tens of milliseconds with detection accuracy up to sub-pixel level, which has good practicability.

关 键 词：极点四边形 矩形拟合 视觉检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]