Research On The Locating Of Cross Mark

Placement machine is the core equipment of SMT production line. Theproduction of placement machine reflects the electronic manufacturing capabilities ofa country or region. China does not have the capability to design and product theplacement machine, and the domestic market is controlled by foreign equipmentmanufacturers. Therefor, the research of the placement machine is very meaningful.The vision system is an important part of the placement machine. The paper bases onthe cross mark positioning in the vision system, and does an research on the templatematching, edge detection, Hough transform, etc. The main contents are:Firstly, the importance and principles of mark positioning is explained. Commontypes of marks are introduced, at the same time the whole idea of positioningalgorithm is made, including extraction of RoI(Region of Interest), edge detection,extraction of lines, find the central location of cross mark, etc.Secondly, a research of the existing image matching methods is did, includingNGC, SSDA, projection matching, etc. The advantages and disadvantages of thesealgorithms are analyzed. Improved projection matching algorithms are proposed.Experiments show that the improved algorithm can effectively improve the efficiencyof image matching.In addition, edge detection techniques are studied. The existing edge detectionmethods are analyzed, including Roberts, Prewitt, Sobel, Laplacian,LoG(Laplacian-of-Gaussian), Canny, etc. The advantage and disadvantage of thesemethods are analyzed through experiments. Improved Canny algorithms are proposed.Experiments show that the improved algorithm can effectively achieve good results inedge detection.Finally, the lines of cross mark are extracted using the Hough transform, and theefficiency of the standard Hough transform is improved according to the knowninformation. The central location of cross mark is found through it’s geometricpositional relationship with the lines. Experimental results show that the algorithm isvery efficient and robust.