Design and Development of Automatic Visual Inspection System for Manufacturing

Abstract

This study is paying special attention to Printed Circuit Boards (PCBs) manufacturing as one of the most important applications for Automatic Visual Inspection (AVI). It provides fast modified procedure to detect the components and the solder-joints of PCBs. In the Automatic Optical Inspection (AOI), one of the AVI approaches, the principal challenges in PCB analysis lie in the high reflectance of the surface, the un-even illumination, the specular nature of the solder joints, and the potential complexity of the background. For industrial application, the feasibility and reliability form two major factors to keep the marketing competition. In case of the PCB, the minimization and the advancing in the PCB manufacturing multiplies the time-consuming for inspection. The suggested methodology employs the Discrete Cosine Transform (DCT) to enhance the desired feature cohesion. Furthermore, the color disturbance is involved to illustrate the components. Finally, the detection has been done by employing a multi-stage segmentation. The classifier is designed to detect five different classes depicts the soldering conditions: good, missing, no-solder, exceed and bridged. The features of four different classes are extracted using a time-frequency localization wavelet. The experiment expresses good results supported by the near optimality of the Log-Gabor filter bank. This allows the image to convolve with a Gabor atoms in the logarithmic prospective, whereas a zero DC component. The obtained results were undergone upon a fast time implemented algorithm in order to meet the flexibility conditions