Research On The Key Technology Of Vision Detection Of High Precision SMT System

The visual inspection of the placement machine is a key part of the placement machine system in the placement machine.Its key technologies include: position estimation of the placement part,and positioning of the PCB reference point.Therefore,it has the characteristics of great research significance and high challenge.Because of non-standard features such as different types of shaped parts,and large differences in size and appearance,traditional methods for visually detecting shaped parts usually need to design matching algorithms for each shaped part.Additionally,with the development of PCB toward miniaturization and high integration,the demand for precision and speed of PCB reference point detection by the placement machine has also increased.There is room for improvement in the accuracy and speed of existing algorithms.Therefore,this paper investigate the functional requirements of the placement machine and the status of research at home and abroad deeply,analyzing the key technologies and difficulties in the visual inspection of automatic SMT machines.Lastly,clarifying the research significance and content of this article.In addition,this paper focuses on the problem of pose estimation and PCB mark points location.The paper main research content includes four parts:1.The function of image filtering algorithm and image threshold segmentation algorithm in image preprocessing of SMT is analyzed and the classical algorithm principle and implementation steps are briefly summarized in this part.In addition,through the algorithm experimental comparison of test images actually taken in the SMT production line,the most appropriate filtering algorithm and threshold segmentation algorithm in SMT visual detection are selected according to the processing effect and time consumption of the algorithm to complete the image denoising and binary segmentation.2.In this part,a template matching algorithm based on edge gradient information is designed to estimate the position and pose of the special-shaped patch elements and solve the problem of rectifying the deviation of the special-shaped patch.Based on the edge gradient information,the template matching algorithm uses the gradient vector of the contour point of the patch element as the matching information.The author selects a stable,reliable andconvenient calculation of similarity measure function,extracting the edge gradient information quickly as characteristic information,designing of matching simple way,setting up appropriate similarity measures to stop the standard and the acceleration of using pyramid algorithm guarantee the running time of the algorithm,finally achieving abnormity patch element rotation angle and position location.3.Aiming at the problem of PCB reference point positioning,considering that PCB circular reference points have higher requirements for the quality of the circle and that the circular reference points will become elliptical rather than ideally circular after imaging,this article uses the curve fitting principle to extract subpixel edges point,and the ellipse fitting method based on the least square method is used instead of the traditional circle fitting method.This paper proposes a PCB reference point positioning algorithm based on sub-pixels,which improves the positioning accuracy and robustness of PCB reference points,and solves the problem of inaccurate positioning or inability to locate due to damaged reference points.4.The author uses the C ++ programming language to implement algorithms and software interfaces in the Visual Studio 2015 environment,and Realize the function of correction of irregular shaped parts and fast positioning of PCB reference points.Finally,the actual performance of the algorithm is evaluated from the three aspects of robustness,accuracy and time-consuming through experiments.