Based On Machine Vision Inspection Of Agricultural Materials Classification System Key Technology

Rice quality inspection process, the grain type of rice is the most intuitivedistinction between rice quality indicators, so a lot of the main indicators of ricequality inspection institutions regard the grain type of rice as the rice qualitytesting. The current low level of agricultural mechanization in China, mainly relyon manual detection of rice grain, although this method easy to implement, but theefficiency is too low, too much subjective factors to rice production and sale of alot of trouble.In this paper, a large grain of rice detection system based on machinevision technology, the detection of rice quality is of great significance.In this paper, work and innovation are as follows:(1) Has designed a large grain of rice based on machine vision technologyidentification devices in this device using the FPGA to control the linear CCDacquisition rice image and then use the AD9821chip and chip AD9823differentialamplifier and analog output analog signal of linear CCDdigital conversion.(2) The gray level transformation, histogram, median filtering, imageenhancement, image pre-processing method of collection to the original image.(3) Extract characteristic parameters of rice area, perimeter, long axis, shortaxis13as the basis of the type identification of a large grain of rice under the ricemorphological characteristics.13characteristic parameters are not mutuallyindependent, they carry rice characteristics of information redundancy, so usingprincipal component analysis extracted characteristic parameters of conductdimensionality reduction and go redundant, and use to generate the three mainingredients instead of rice,13characteristic parameters as the basis of rice grainidentification.(4) BP neural network weights using genetic algorithm optimization of the ricegrain type, and then use the improved BP neural network and the standard BPneural network to identify and compare their superiority. Experimental results showthat: the genetic algorithm improved BP neural network is superior to the standardBP neural network in the rice grain on the recognition accuracy and real-time.