为克服现有自动光学检测（AOI）算法的缺陷,针对AOI的特点,对传统的双向二维线性判别方法进行改进,提出一种基于加权双向二维线性判别方法的焊点检测算法.在计算协方差矩阵时,对不同的类别以及类内不同的样本进行加权,从而提取更有判别力的特征.将改进后的算法应用于焊点检测,可实现对不同批次用料的自动分类.文中提出的算法检测速度可以满足实际需要,且与现有AOI系统中常用的图像对比算法和图像分析法相比,可以显著降低误报率.

In order to overcome the disadvantages of the existing automatic optical inspection（AOI） algorithm for solder joints,a new inspection algorithm based on weighted bidirectional 2D linear discrimination analysis is proposed,which takes into account the characteristics of AOI and improves the conventional bidirectional 2D linear discrimination analysis by weighting different classes and samples during the calculation of covariance matrix.By this algorithm,more discriminative features can be extracted.The improved algorithm is then applied to the solder joint inspection to automatically classify the solders from different batches.Experimental results show that,as compared with the commonly-used algorithm based on image comparison or image analysis,the proposed algorithm significantly reduces the misdiagnosis rate with the detection speed meeting the practical demands well.