The Key Technologies Study Of Image Processing For Chip Mounter

In recent years, with the face of the problems of growing "labor shortage", electronics manufacturers and OEMs of China are being developing and importing large quantities of artificial intelligence equipment to replace manual labor. Machine vision technology, as a new intelligent manufacturing means, is not only the power source of the industrial manufacturing industry, but also the core means to promote the progress of industrial of China. The most prominent feature of machine vision is to replace the traditional manual labor with intelligent automation equipment, thereby increasing the productivity of industrial processes. Particularly in harsh environments or high precision, artificial vision simply cannot meet the production needs of the premise of industrial machine vision, but also demonstrates the occupied important position in industrial intelligent manufacturing. In the SMT industry, the detection success of electronic components SMD placement machine is whether or not high-precision, high-speed placement based on the premise.Thesis studies the basic theory of digital image processing algorithms in visual detection. Through the comparison of the median filter, mean filtering algorithm and so on, some of the traditional filtering algorithms carry out the weighted pixel processing in the form of a sliding window, which is bound to cause image blur and conducive to the subsequent defect detection, and ultimately affect the detection accuracy. Therefore, some of the common experimental analysis of filtering algorithms are done, the experimental results show that user-defined median filtering algorithm not only can remove noise in the image effectively, but also can retain edge detail better, and can in compromise on the speed and effectiveness to meet the high-speed SMT placement needs. By comparing the iterative method, Otsu method and other threshold segmentation methods, that showed Otsu image threshold method strikes closer to the actual value.In the extraction process of key feature points of the image, a general shape template matching and feature-based positioning of the target component caused relatively large position error due to the factors surrounding the device. Therefore, for the mark points of PCB, this article proposed an algorithm based on the circle of the ellipse to quickly locate a reference point mark, while for SMD devices, adopted the shape-based matching algorithm for secondary relocations chip components to locate and follow-up defects.In the identification process of chip components, the different degrees of oxidation of metal of IC pins, is bound to cause the pin fracture when the camera captures the images, so it can’t be properly used to detect IC appearance. Therefore, this paper analyzed the performance indicators needed in defect detection of chip mounter, came up with a comprehensive repair techniques and defect detection algorithms, in actual industrial production process through a long-term stable performance testing, and achieved certain economic benefits. Finally, the content of the research work is summarized, and the future of work is prospected.