การตรวจสอบข้อบกพร่องแผ่นวงจรพิมพ์ด้วยเครื่องมือทางการประมวลผลภาพ

Abstract

Printed circuit board (PCB) inspection systems are essential in the electronics manufacturing industry. Designing an effective inspection system will not only help to get good quality PCBs but it also speeds up the production process and results in lower production costs. Although existing PCB inspection technology is capable of detecting defects, not all defects can be classified accurately and comprehensively. This research therefore proposes a PCB flaw detection system that can detect defects. It identifies the location, size, shape and nature of the defect and can identify 1 4 types of defects on a single-sided blank PCB with a classification error of no more than 10%. This research uses fuzzy c-means clustering for image segmentation before defect detection via image subtraction. Arithmetic and logic operations, circle hough transform (CHT), morphological reconstruction (MR), and connected component labeling (CCL) are used in defect classification. The research results showed that the proposed PCB flaw inspection system is 100% effective in detecting defects with the ability to clearly identify the location, size, shape, and nature of defects. All 14-known PCB defect types can also be classified, with an average PCB defect classification accuracy of about 99.05%.