基于点模式的BGA芯片定位算法研究

Research on Image Locating Algorithm with BGA Package Based on Point Pattern Match

该文针对贴片机生产中BGA封装芯片的中心定位问题,分析了现有模板匹配算法的不足,提出了基于点模式的快速定位算法。该算法针对BGA芯片引脚分布对称的特点,提出改进的点模式识别算法。首先由BGA芯片封装的管脚定义生成0°的管脚模板坐标集,用B lob点分析算法得到目标图象管脚坐标点集并取x轴坐标最小的一点作为启始匹配点,以该点为原点求取目标点集与模板点集的极坐标仿射变换参数,再对该仿射变换参数作可靠性验证。该算法大大减少了匹配的运算量。实验效果表明该算法的识别速度和精度都能满足实际生产的需要,并具有较强的鲁棒性。

Aimed at the image locating problem with BGA package,this article analyses the shortage of general pattern match algorithm then presents a fast locating algorithm based on point pattern match method. In view of the symmetrical distributing of the BGA ball grid,the article proposes an improved point pattern match algorithm. First it generates the pattern set of 0 （ from the BGA package definition, then gets the object set of the image using the Blob analysis algorithm and lets the point which has the minimal X coordinate to be the first match point. Using this point as reference origin and calculates the parameters of the affine transformation between object set and pattern set, then validates the reliability of the parameters. The algorithm reduces the time - cost operation greatly. The result shows that the locating speed and the accuracy of the algorithm are well and can meet the requirement of practical application. The algorithm also has strong robust.