Research On Defect Detection Method Of Flexible Printed Circuit Board

With the rapid development of industrial technology, FPC(Flexible Printed Circuit boards) is widely used in aerospace, military, mobile communication and consumable fields with its characters of light-weight, thin-thickness and free- bending. However, the limitation of production technology and craft skills makes the FPC board of industrial field exist quality problems in general. The traditional artificial detection method with high cost, low efficiency, and subjective consciousness,which was difficult to guarantee the FPC quality.Therefore,it is urgent to develop a method with fast and efficient characteristics. Vision detection technology based on image processing with its non-contact, high-intelligence and high-efficiency, has become a inevitable trend of industrial development.Compared with the conventional PCB, FPC board with flexible materials is prone to reasonable distortion. Therefore, defect detection algorithms for PCB can not be directly applied in the process of FPC defect detection,and it need to develop a special quality detection method for FPC. In this paper, quality of all kinds of keyboard FPC for industrial production line were inspected. According to the image data which had been transferred to the computer memory, line defects(open circuit, short circuit, bulge, indentation, et al.) were extracted and display based on image processing technology. The research of the paper was mainly focused on the following aspects:(1) On the basis of existing algorithms on printed circuit board detection, According to the ideology of template matching, the paper firstly researched methods of FPC defect detection from the global matching aspects. For the FPC image distortion, the image can not completely be matched,studied the method of FPC defect detection from the local matching aspects.(2) Considering the insufficiency of template matching method, methods of FPC defect detection based on line features: detection method based on contour features of lines firstly was researched, which is not conducive to industrial automation production because speed of the algorithm was too slow by the impact of image format. Secondly, another defect detection method based on endpoints and junction points information of image skeleton was studied, experiments show that the method can quickly locate defective information of rule lines, but for more complex irregular lines, such as LED due to the inability to effectivelyinterpret contour lines,the method appeared misjudgment.(3) For the problem that both template matching method and image skeleton method can not realize defect detection of the whole circuit board. The comprehensive consideration of the template matching method and line features extraction, A hybrid detection method based on the combination of image skeleton and template matching was proposed.According to the above algorithm, the FPC defect detection system was complete with C++software language in Windows operating system. The system totally includes image acquisition control, defect detection and processing, data communication control and other functional modules.A large number of real-time printing detection experiments for the FPC substrates with the size of 680mm×450mm(about 16384×24285 pixels) were conducted, the algorithm defect detection types and average detection speed can meet the quality requirements of quality inspection, which can be applied in the industrial field.