Automatic Detection System of Surface Defects on Metal Film Resistors Based on Machine Vision

An automatic detection system of surface defects based on machine vision was developed. Metal film resistors were chosen as model materials. A CCD camera captures three images while the metal film resistor rotate once to get its full surface image. Using histogram equalization and median filter, the image contrast was improved and contours were smoothed. Global threshold segmentation distinguished the band from the background. Then the anchor point was found in the upper-left corner of the resistor. Band areas were extracted according to the anchor point. In situations where band areas were found, proposed algorithm recognized defects in terms of the number of band contours and the times from black to white in each pixel-width line. Depending on the above results, PLC controls the motion of the relay, implementing the classification of qualified resistors and defective resistors. The system was tested to detect defects in different surroundings and showed a success rate of more than 95%. Future work is being done to design an automatic feeding mechanism to improve feeding speed.