Research And Exploitation Of The High-speed Screen Printing Machine Control System Based On Computer Vision Orientation

With the development of science and technology, smart phones and tablet PCs areincreasingly favored by consumers, while the capacitive touch screen is a key component ofthese two products. In the production of capacitive touch screen, graphic screen printing is animportant process and also difficult process. It needs to print graphic in the correct position ofthe screen glass, and position accuracy is required highly. Positional deviation is requiredwithin±0.03mm by large companies like Apple, Black Berry etc. The difficulty is thatworker can’t place the glass to such a precise location. In addition, sizes of pieces are notunitizing, with tolerance±0.05mm. However, it is required to print the graphic in the centerof the glass no matter it is big or small.According to the above requirement, a high speed screen printing machine controlsystem based on visual orientation is designed by using the current printing machine as aplant. This system adopts4CCDs to capture the edge image of glass. It computes thedeviation between current position and reference position by image recognition and locationalgorithm; then drives the adjusting platform to precisely adjust the position of glass; at last,controls the other mechanism to complete screen printing and glass unloading.The hardware of control system adopts PC as a host controller, fiber optical ring networkas communication mode to provide control interface. And the software is realized bymodifying the open Linux CNC system software and doing some upper development. Firstly,the former single channel CNC is expanded into dual-channel CNC in the entironment ofreal-time Linux OS. Secondly, two NC code programs of dual-channel are designed to realizeto all the actions of the screen printing machine. Thirdly, the image capturing module isdesigned by V4L2interface. Finally, GTK+library is used to create graphic user interface.The technical cores of the whole system software are visual recognition and positioningalgorithm. In image recognition, image filtering, edge recognition, sub-pixel location, andLeast Squares are use to get the precise line equation form the image captured by CCD. Andin positioning algorithm, degree of freedom is analysed to prove the glass positon can befigure up only by4CCDs, and then plane geometry and progression approximate method areapplyed to derive the platical localization algorithm.The control system has passed trial run on a real machine. It can achieve the entirecontrol task and all the technological requirements. Entire process including positionadjusting, printing and glass unloading can be completed in less than10seconds.