BARE BOARD MULTILAYER PRINTED CIRCUIT BOARD DEFECT IDENTIFICATION FOLLOWED BY LOCALIZATION

Abstract

In recent years, the Multilayer Printed Circuit Boards are becoming the mainstream of all electronic devices due to its various benefits such as high assembly compactness, small size, and worthy quality. Hence manufacturing of defect-free Multilayer Printed Circuit Board has gained a lot of importance. This paper is mainly focusing on the identification and localization of most commonly occurring defects like Trace Cut, Pad Injury and Trace Short. In this work, a non-contact and referential based algorithm is developed for the inspection of real Multilayer Printed Circuit Boards defects. In this approach, firstly the RGB image of both reference image which is having no defect and Test image will be split in to individual channels of R, G and B. The segmentation process will be applied on individual channels of both reference and test image to obtain the binary images. The comparison of binary images of individual channels is performed to obtain three difference images. The three difference images will indicate Pad Injury, Trace cut and Trace short defects respectively. Once the defects are identified the location of the individual type of defect is obtained by Difference of Gaussian method and maximum filter. Finally, the location of the defect will be highlighted along with label on the test PCB image. The experimentation shows that, the proposed approach successfully performs the inspection of multilayer boards having single defect, multiple similar defect and multiple different defects. Time taken for the inspection of the board is presented and it is ranging from 1627 to 1659 milliseconds depending on the dimension and number of faults. Extensive comparisons of the proposed method with previous methods demonstrate that our proposed approach is more feasible and effective for the inspection of Multilayer PCB for Pad Injury, Trace cut and Trace Short defects.