Vision System’s Research Based On OpenCV In The SMM Of SMT

The research platform in this paper is QSP-2Surface Mounting Machine in SMT Laboratory of the Electrical and Mechanical Engineering College in Soochow University. On the basis of the original equipment, this paper upgraded part of the hardware control system, developed a SMT corrective patch visual system based on OpenCV, and control the Mounter patch process with the visual image processing function in the OpenCV Library. After several debuggings and modifications, it meets the requirements of speed and accuracy.The visual system in this paper designed and researched mainly included two parts. Firstly, It realized the PCB positioning:after the PCB fixed its placement, the camera driven by the holder obtained a reference point image, operated the image with some image processing including median filtering、closing operation and so on. Used Hough circle transform to detect the circle image coordinate of Mark point on the board, transformed it into machine coordinate, calculated the deflection angle of the PCB using diagonal two Mark point information. Then used one Mark circle as a reference point, calculated the actual machine position of the elements on the board and stored in the computer based on the component information in the database. Secondly, fulfilled the correction of the patch components:carried out linear fitting of the component side using the breadth-first algorithm and the method of least squares, calculated the deflection angle of the element, computed element center point taking advantage of all the contour center points, and calculated the deflection distance of the central point, sent the deflection angle and deflection value of the center point to the executing agency for corrective compensation patch.