Abstract:

For the problem of SMT components pin welding online detection in PCB, a pin welding defect detection based on machine vision algorithm is proposed. To achieve automatic identification, this paper extracts the features of pin, applies area method and the connected domain method comprehensively to recognize patch pin attached solder ball, foot bridge, lack of welding and qualified products. Therefore, an approach of adaptive correction of parameter threshold based on statistical method is put forward, which can solve the problem of limitation of the fixed threshold effectively and make the algorithm apply to the environment with different light intensity. The experimental results show that the algorithm of PCB in detecting welding defects in SMT chip pins has higher accuracy.

Key words: pins of SMT; welding detection; feature extraction; machine vision