Study On Vision Image Grabbing System Based On DSP And Character Recognition Method For Chinese Chess Playing Robot

Under the hold of the Excellent Talents Project of Beijing City, a Chinese chess playing robot is being researched at the Group of Robot of the school of Mech-electronical Engineering of NCUT. This robot was designed to accomplish the whole process of chess-playing by itself, including of recognizing chessboard, confirming chess-manual and moving chessman. It was comprised by Vision System, Management System, Manipulator System and Analysis System.In this paper, the Vision System was chosen as the goal of research and some technology contents were studied.First, due to the unique advantages, a vision-image grabbing path by the way of DSP (Digital Signal Processor) was proposed. DM642was chosen and the grabbing system was built up. By it, the original image of the chessboard was collected.The technology of imaging pre-processing was applied for graying, filtering, edging and binarization of the original images.Then, the method of character recognition and its application were concluded and the characteristics of chess character recognition were analyzed. The method of character recognition based on annual ring was chosen for our robot project. An algorithmic of backup eigenvalues was proposed. The algorithmic flow and the program of this method were designed. It is proved that this method is accurate and has a high recognition rate.Thirdly, the two-dimension coordinate system was adopted to locate the chessmen. The method of chess-orientation by the inspections of chessboard grid crossing and its local area was proposed for the location of chessman. The preliminary program was designed.These hardwork system, software algorithmic and programme would be a part of the finially whole vision system of this robot.