Computer Vision Inspection And Classification On Printed Circuit Boards For Flux Defects

• Main Author: Ang, Teoh Ong
• Format: Thesis
• Language: English; English
• Published: UTeM 2016
• Subjects: T Technology (General); TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utem.edu.my/id/eprint/18521/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=100384; HF5351.M34 2016

The manual inspection of Printed Circuit Boards (PCB) is labor intensive and slow down the production line. During the assembly process, the defective PCBs with flux defects if not detected and remove, it can create corrosion and cause harmful effects on the board itself. As such, an automated inspection system is very much needed to overcome the aforementioned problems in PCB production line. The main objective of this work is to develop a real-time machine vision system for quality assessment of PCBs by detecting defectives PCBs. The proposed system should be able to detect flux defect on PCB board during the re-flow process and achieve good accuracy of the PCB quality checking. The proposed system is named as An Automatic Inspection System for Printed Circuit Boards (AIS-PCB), involves design and fabrication of a total automation control system involving the use of mechanical PCB loader/un-loader, robotic pneumatic system handler with vacuum cap and a vision inspection station that makes a decision either to accept or reject. The decision making part involves classifier training of PCB images. Prior to ANN training, the images need to be processed by the image processing and feature extraction. The image processing system is based on pattern matching and color image analysis techniques. The shape of the PCB pins is analyzed by using pattern matching technique to detect the PCB flux defect area. After that, the color analysis of the flux defect on a PCB boards are processed based on their red color pixel percentage in Red, Green and Blue (RGB) model. The red color filter band mean value of histogram is measured and compared to the value threshold to determine the occurrence of flux defect on the PCBs. The texture of the PCB flux defect can also be extracted based on line detection of the gradient field PCB images and feature indexing by using Radon transform-based approach. The feed-forward back-propagation (FFBP) model is used as classifier to classify the product quality of the PCBs via a learning concept. A number of trainings using the FFBP are performed for the classifier to learn and match the targets. The learned classifier, when tested on the PCBs from a factory’s production line, achieves a grading accuracy of coefficient of efficiency (COE) greater than 95%. As such, it can be concluded that the developed AIS-PCB system has shown promising results by successfully classifying flux defects in PCBs through visual information and facilitates automatic inspection, thereby aiding humans in conducting rapid inspections.