The Research On Head Milled Rice Detection And Recognition System

Head milled rice rate is an important national parameter in out country's rice purchase and transaction, and is not realized by machine selection. The national requirement of head milled rice automatic detection system is quick and accurate, the time of recognition is below 10ms. The recognition's key is diameter feature extraction of rice longitudinal section. Only using FPGA technology can ensure that the system is real-time and accurate.In response to this issue, we design the head milled rice dynamic detection system. The system includes designing of hardware circuit, designing of software and algorithm. It stands at the core position of XILINX Sparten3 series XC3S200 FPGA chip, CPLD chip XC9572XL and high-speed linear array CCD sensors TCD1209. The falling rice dynamic image is obtained by CCD sensor, then features are extracted. By communication with host computer ARM9 system, finally it achieves to sort qualified and unqualified rice and calculate rice qualified indexes such as the head rice milling rate. Head milled rice dynamic detection system hardware circuit contains three independent modules; image acquisition circuit board, image process circuit board and gun signal driving board. The separate design of sampling circuit and process circuit avoid line interference, and is helpful to system modularization.The system takes ISE8.2 for compiling and exploiting environment, which implements the driving for TCD1209, the access of Block RAM image, the detection algorithm and the accurate location of gun signal system. Four algorithms are designed for extracting feature parameter, which include pixels number, center-edge rotation search, MER recognition and parameter matching, and the optimum algorithm is determined by comparing each speed and accuracy. By whole machine debugging, the system can accurately extract image parameters and get head milling rice rate calculation.