Research On Coal And Gangue Sorting System Based On Davinci Video Image Technology

Coal preparation is an important part of clean coal technology. The traditional coal and gangue preparation usually uses wet cleaning method and dry method which are based on the physical principle, and these methods, especially the wet cleaning method(HM method and jigging method), waste the water resource seriously and pollute the environment obviously. In the recent years, there are some coal preparation research about image processing. The TI company promoted the DaVinci technology which is designed for a new solution of digital video in2005. The DaVinci technology can realize complex image processing algorithm and can process the video in the real time, at the same time, it makes the exploiting procedure easier, and compared to the computer, the DaVinci technology provides the video application with less cost and more portable equipment, so in this paper, we propose the method of coal and gangue preparation which takes the DaVinci video image processing technology as the core. In this system, we choose DaVinci processor DM6437. We take the DM6437as the core, and design the hardware circuit based on the function and demand of system.The sorting system’s chief work is to recognize the real-time coal and gangue, and to hit the gangue by the hit device, meanwhile, it makes statistics about the number of the coal and gangue. In this research, we build the structure of sorting system and do the image processing research algorithm of recognition and count, and in the research of algorithm and program, we protrude the real-time in the video processing. As we mainly use the difference of gray value between coal and gangue in the recognition, we do a series of experiments to find out the gray value interval of coal and gangue. In the image segmentation, we present the specific background subtraction method to extract the foreground, which uses the difference between coal and gangue and belt, and we use the method of Ostu and double peak to determine the best threshold, because most of the coal will show bright pot in the daylight lamp irradiation, so after the extraction of foreground, we need use the morphological processing algorithms to do further processing in order to scatter the bright spot, at the same time, the area of gangue becomes more obvious, which is easier to recognize the coal and gangue. In the last, the control signal is passed to PLC, then the PLC control the high-pressure valve to hit the gangue. After we use this method, the accuracy of recognition for belt movement of coal and gangue can reach more than90%. The traditional coal and gangue counting method by the photoelectric has serious defects, affecting the accuracy of the sorting system and the evaluation of coal quality. In this paper, we propose the use of the continuous video image to reflect the characteristic of coal and gangue outer contour, and we adopt some image processing methods to count the coal and gangue with the coal and gangue’s contour change. The intelligent method substantially improves the accuracy of the count.