Image Processing and Machine Learning for the Detection of Defects in PCB Images

The presented work considers three approaches to detecting defects in printed circuit boards. Each combines two components: algorithmic and software for image processing and machine learning based on image features identified in the first step. The approaches are as follows: division of boards into correct and defective boards without indicating the types of defects, determination of connection defects and redundant or missing components, and determination of defectiveness of tracks and contacts by analyzing individual components of the board. Color manipulation algorithms and additional image processing tools have been developed. Neural networks of artificial intelligence, which are available on the Internet, were used in the work.