| Chip mounter is the core equipment of SMT production line, which is monopolized byforeign producers.So it is important to achieve the taget of producing chip mounters of ournation. In this paper, key technologies of image processing of the surface mount componentsfor chip mounter is researched, including image filtering, image thresholding segmentation，edge detection and Image position and angle detection. Specific implementation methods areprobed, such as image acquisition, image processing, position recognition algorithm andsystem software designing and programming. The main work consists of three parts:Firstly, the key technologies of chip mounter are analysied, and the system architectureof chip mounter is planned. And the the surface mount components are classified and themounting acceptance standards are discussed. And the requirements of the system areproposed and the hardware and software structure of the system are planned, including imageprocessing module, light source controlling module, database management module andmechanical module.Secondly, the key technologies of image processing for the chip mounter are mainlyresearched on. For different surface mount components, a variety of image pre-processingapproaches are used, such as image filtering, thresholding segmentation and edge detection,and experiments proof that Mean filter and Canny edge detection operator are fitted for theimage pre-processing of the images of surface mount components. Agaist interference andedge missing when detecting the edges of the images, the edge-repair method based on theminimum bounding rectangle is proposed. Algorithms for the position and angle detection ofChip class, SOP and QFP class are also proposed. Agaist the Chip class and SOP and QFPclass, position and angle detection algorithm based on the image centroid fitting andleast-squares fitting are proposed. And experiments proof the algorithms achieve the repeatedaccuracy that is about50um.In addition, the identification of mark points of the PCB is researched on. Andalgorithms against round shape point base on fitting center of a figure and Hough transform totarget the center are proposed. Agaist mark points of other shapes, algorithm based on SIFT isproposed. Through experiments, the algorithm has a high positioning accuracy.Finally, the optimization of the mounting paths based on ant colony algorithm isdiscussed about, which improves the efficiency of the chip mounter. The database structureand the light source adjusting algorithm are analysied; process for software programs issummarized. Based on the analysis of the system software architecture and functional modules, the sofeware is programmed by the Windows platform using MFC. The applicationresults show that the related technical is effective, and it is a great significance to improve thepatch level of the electronics industry. |
