Research On Image Registration Technology And Application In PCB Manufacturing

In recent years, with the continuous development of machine vision technology andink-jet printing technology, PCB (Printed Circuit Board, Printed Circuit Board) industryushered in new opportunities for development. Although after one hundred years ofdevelopment and reform, traditional printed circuit board manufacturing process andtechnology has been very mature, but also facing many problems, such as the workingprocedure is various, complicated, low efficiency, serious pollution, costly, and manyother issues. and the PCB printing. Just to make up for the defects, and fundamentallyovercomes the traditional PCB manufacturing process and technology problems. PCB inkprinter is a perfect solution to make up for all these defects, and fundamentally overcomesthe traditional PCB manufacturing process and technology problems. Multi-layer, highdensity, high reliability, miniaturization and pollution-free green products is thedevelopment direction of printed circuit board in the future. So, with the call of The Times,develop a highly efficient, energy saving, intelligent, green pollution-free PCB printer hasbecome a pressing task with practical significance.In this paper, we adopt advanced technology of machine vision, digital jet printingtechnology and mechatronics technology, developed a set of PCB printer smart registrationsystem, and give the system schematic diagram of structure and the process details andmethods of the system. The system includes motion control module, image acquisitionmodule, intelligent registration module, etc. Registration system use template matching tocomplete the registration, because in the application of PCB registration,it use round holeas PCB mark, so the PCB mark has the rotational invariance, so even if the PCB has thepresence of rotation, we can still detect the PCB mark. This is the theory basis of why weuse template matching for registration in the system. We get the three template pictures ofregistration hole when the first time people do register, later on, we can use the threetemplate pictures of registration hole and the design positions of the three registration holeto calculate the real position on the PCB board. It will reduce the time to align the PCBboard manually, also increase the automaticity of the system. To increase the robustness ofthe system, we add human intervention when automatic align system fail.In terms of theoretical innovation, in view of the existing weakness in real-time imageregistration technology, we first pioneering propose five corner point features, also we give the calculation method of the5corner point features. We use SUSAN operator for cornerdetection and edge detection all at once by adjust the threshold used in the SUSANoperator. In the end, we show how to use the5corner point features in image registration,and the result show it can reduce the calculated quantity and improve the speed of imageregistration.