Occlusion Element Matching Algorithm Based On The Best Similarity Point Pair

With the increasing integration of electronic components,it is more difficult to detect defects in Printed Circuit boards(PCB).When Automated Optical Inspection(AOI)detects the defects of components in PCB,the components should be firstly positioned accurately.Because the components are not soldered during defect detection,and there is vibration in the transmission process of plug-in assembly line,which leads to the occlusion of some components,the success rate of component matching will be reduced,and the accuracy of AOI defect detection will be affected eventually.When the components have occlusion,it will affect the accuracy of AOI in defect detection of components.In this paper,the corresponding research on this problem is carried out and the solution is proposed.Through theoretical analysis and experimental verification of the traditional image matching algorithm,it is found that the components cannot be matched accurately when there is occlusion.The algorithm of best-buddies similarity(BBS)can not only accurately match the unblocked images of the target,but also accurately match the blocked images of the target with good robustness.By studying the core idea of BBS algorithm and carrying out experiments,it is found that this algorithm matches various defects of occlusion components.Aiming at these defects and combining with the characteristics of PCB,an improved algorithm is proposed.Firstly,the integral graph is used to roughly match the target image to obtain the similar region of the target image and the template image.Only the best similarity pair of the target image sub-graph located in the similar region in the upper left corner is calculated to reduce the algorithm time.Then,combined with the characteristics of PCB components,gradient information was added into the distance function of BBS algorithm,and the calculation method of spatial distance was changed to Manhattan distance.In order to reduce noise interference,pixel blocks were desiccated.Finally,the concept of weight is proposed to eliminate the contingency problem in the matching process.The improved processing of BBS algorithm can not only match the unoccluded components,but also match the occluded components accurately and in real time.No matter the occlusion exists or not,the improved BBS algorithm in this paper can match the components accurately in real time.When components are accurately matched,in order toreduce the influence of shielding region on AOI defect detection and improve the accuracy of AOI defect detection,it is necessary to judge whether the matched components have shielding or not.In this paper,the mesh density clustering algorithm is used to cluster the best similarity pairs and locate the partially occluded area.The occluded area is calculated to determine whether the components have occlusion or not.When the components are not occluded,AOI directly detects the defects of the components.When the components have occlusion,this paper uses the scanning line seed filling algorithm to search the occlusion area,accurately locate the occlusion area of components,and eliminate the influence of occlusion area on AOI defect detection.In order to verify the real-time and matching accuracy of this algorithm,this algorithm is compared with traditional image matching algorithm and BBS algorithm.From the perspective of real-time analysis,although the algorithm in this paper takes longer time than the traditional image matching algorithm,compared with the BBS algorithm,the time is significantly reduced,which can meet the real-time requirements.From the analysis of matching accuracy,when the components are not blocked,the algorithm and BBS algorithm can accurately match the components.When the components have occlusion,the matching accuracy and robustness of the proposed algorithm are higher than that of the BBS algorithm.Finally,the robustness of the algorithm proposed in this paper is verified.For different components with occlusion,the algorithm proposed in this paper can accurately locate the occlusion region,reduce the influence of occlusion region on AOI defect detection,and improve the accuracy of AOI in defect detection of occlusion components.