Research On Some Key Technologies Of PCB Automatic Optical Inspection Equipment On Array-camera Arrayed Based On Watershed Algorithm

Machine Vision is a device which automatic receives and processes the real picture through a serious of appliances which conclude the optical instruments and the non-contact sensors.Also it is used to get valid information or control the robot motion.The paper introduces the machine vision applied in AOI(Automatic Optic Inspection)which is used in the rigid circuit board defect detecting in the aspects of abstract,project research,key technology and the stuff.Contrasting with the development of the related technologies,we propose the research planning,key points and the directions,meanwhile we have got two important indicators including the detection accuracy and detection efficiency.Comparing with the applied detecting methods about the accuracy and the efficiency in automatic optic inspection,we make new design the hardware structure and the optimization algorithm considering with the practical application experiences in order to control the accuracy and the efficiency without any more costing.Choosing the right things is most important for the system,for example light source,camera lens,camera,hardware platform and the stuff.The paper proposes a new way,multi-camera shooting in a scheduled array,instead of the normal line scan CCD camera.We take measure of a single exposure to obtain an image of a large format PCB substrate.Then we take processing of the image through the improved picture processing algorithm for example local threshold processing,subpixel edge detection,camera calibration,image mosaic,image segmentation,defect recognition,etc.The innovation method based on the traditional method is to optimize the hardware instruments and the structure and the better algorithm which can be used to improve the system's robustness performance in the paper in order to form an integral detecting system.PCB(Printed Circuit Board)automatic optic inspection device is based on machine vision combining with the customized demands take advantage of the appropriate technology to recognize the defect through the different geometrical characteristic in the image.The multi-camera shooting method can get the large-format image about the image to be detected and hold all the information with the way of a single exposure.The way how to get the same resolution image can lower the hardware costing,reduce the time cost and improve the efficiency.At the aspect of the algorithm research,we use the improved-watershed algorithm to get the details about the type of detect and the location.Comparing with the substraction algorithm,the method in this paper can process the marked pixel region instead of the whole region avoiding wasting time.At the same time we can enhance the accuracy during the defect repairing because we can make sure threshold value in the progress of the detection that the substraction can not.It has a stronger robustness and it is convenient to get the detect information and quality improvement with considering the detecting environment and improving the schedule.In the software,we have programed our own software interface and completed the main surface.Also we make an analysis of the Gerber files and images mosaic.At last we store the results of the detecting operation and display them in a form.