X-Ray Imaging Inspection System for Blind Holes in the Intermediate Layer of Printed Circuit Boards with Neural Network Identification

Abstract

Abstract This study presented an X-ray imaging inspection system with a backpropagation neural network that could increase the accuracy of defect detection and classification of blind holes in the intermediate layer of printed circuit boards (PCBs). In this system, a multilayer PCB image was obtained from an X-ray camera. The original image was then converted into a binary image with a noise-suppression filter, and the edge-detection method was used to compare the image with a standard sample. Drilling was based on the hole-position's accuracy measurement to obtain the hole flak figure, which was useful for calculating the drilling coordinate error with a backpropagation neural network. The proposed method could determine the information of the PCB edge test holes automatically. The accuracy of the feature extraction was increased by using the proposed module-detection method, together with image processing and the backpropagation networks process.