Abstract: To improve the accuracy and real-time performance of object region detection in Printed Circuit Board Assembly (PCBA),a PCBA template matching algorithm combined with Gaussian pyramid transformation and new particle swarm optimization algorithm was proposed.The inversed Sigmod function was used to adjust the inertia weight during particle swarm iterating.Adaptive learning factor models of individual and group were constructed respectively.The particle would be adjusted by random momentum factor when it was trapped in local solution under the adaptive criterion.The originate image and the template image were transformed according to Gaussian pyramid with four layers.Coarse matching region of top layer of sub-image was found by proposed particle swarm optimization algorithm,which would generate a neighboring region by inverse transformation of Gaussian pyramid.The neighboring region was compared with corresponding template sub-image,and the matching result was obtained in the bottom layer.The experimental results showed that the proposed method had obviously accuracy and real-time performance in the application of PCBA template matching.

Key words: template matching, Gaussian pyramid transformation, new particle swarm optimization algorithm, adaptive learning factor, printed circuit board assembly

摘要： 为提高印刷电路板装配(PCBA)中目标区域检测的准确性和实时性,提出一种高斯金字塔变换与新粒子群优化算法结合的PCBA模板匹配算法。采用倒Sigmod函数调整粒子群迭代的惯性权值;分别构建个体和群体的自适应学习因子模型;提出粒子是否陷入局部解的自适应判据并对其采用随机动量因子进行调整,从而提出一种新的粒子群优化算法。分别对待匹配图像和模板图像进行4层高斯金字塔变换,采用新粒子群优化算法搜索待匹配顶层子图的粗匹配区域,该区域经高斯金字塔反变换后生成的邻域范围与对应的模板子图进行遍历匹配,在最底层得到最终匹配结果。对比实验结果表明,所提方法在PCBA模板匹配应用中具有准确性和实时性。

关键词: 模板匹配, 高斯金字塔变换, 新粒子群优化算法, 自适应学习因子, 印刷电路板装配