Appearance Of Pcb Image Registration Algorithm Of Photoelectric Detection System Research

Along with the development of the industrial society, the demand of printed circuit board in all walks is increasing. However, due to the limitations of using traditional manual means to detect the printed circuit board (PCB) appearance so as to determine the quality of products, the development of the PCB’s application is limited. The means can neither match the speed of machinery manufacturing, nor meet the high precision demand of industrial production. As a substitute for manual means, PCB appearance photoelectric inspection system is important testing equipment which has been widely used in PCB industry. One of the system’s key technologies is image registration which directly influences the accuracy and efficiency of detection.In this dissertation, the registration algorithms’research and analysis which used by PCB appearance photoelectric detection system were focused on. At the same time, the principle, the process and the performance of the system were explained in detail.Firstly, compared with the traditional manual detection methods, the practicality and the progressive significance of using digital image processing technology in automatic detection were introduced. Secondly, the system’s image registration algorithm was analyzed in this dissertation, and several common registration algorithms’advantages and disadvantages and application occasion were discussed. According to the characteristics of the images of PCB appearance photoelectric inspection system, the images have a large quantity of data, most of the imaging zones have rules appearance, and the characteristics of the color features obvious, so that the image registration algorithm based on the LAB color space and connected components features was put forward. The contents of color space layered, connected components detection, affine transformation, parallel computing and so on were integrated by the algorithm. The algorithm doesn’t need positioning holes, it could register images in any angle and the steps of selecting positioning hole for registration manually could be avoided so that the highly automated production process could be realized. Compared with the traditional method about gray images’feature matching, the method designed in this dissertation has higher precision. Finally, based on the development of the current hardware, the next research content was analyzed, and the direction of the algorithm’s further optimization was proposed.To our knowledge, the method of applying the LAB color space to layered entire image of PCB and then combining the characteristics of the connected components to register images in the PCB appearance photoelectric inspection system has not been reported in China.