Research on the Recognition Algorithm of Circuit Board Welding Defects Based on Machine Vision

To improve the defect detection ability of circuit boards and reduce the missed detection rate and false detection rate, a circuit board welding defect recognition algorithm based on machine vision is proposed. The system obtains the grayscale image of the circuit board to be tested through X-ray source, image intensifier and a Charge Coupled Device (CCD). Noise suppression is performed on all test images using a cumulative sampling noise reduction algorithm. The defect recognition algorithm is realized by using a standard template matching model with multi-angle image acquisition. By setting the best template matching parameter (BTM), the difference area extraction between the test image and the standard image is completed. Then, the calibration transformation of different perspectives is used to complete the iteration of the feature information of the defect area, and the ability of defect detection and identification is improved. The experiment is tested on 15 circuit board images with different types of defects. The results show that the missed detection rates of this algorithm for bridge defects, eccentric defects and solder joint bubble defects are 0.58%, 1.18%, 1.95%, and the false detection rates were 0.12%, 0.86%, 2.34%, respectively. It is significantly better than traditional algorithms. In terms of processing speed and maximum fitness, this algorithm is also slightly better than the two traditional algorithms. In conclusion, this algorithm can better complete the rapid identification of circuit board defect locations.