Quality Grade-Testing of Peanut Based on Image Processing

The quality of peanut kernels is referred to the every aspect of the profit of supply and marketing. A BP neural network model of quality grade testing and identification is built which is based on 52 appearance features such as the form, texture, and color and so on with technology of computer image processing. The testing aiming at 1400 grains is made separately in unsound kernel, mildewing, impurity, hetero-variety and other aspects with the result of the correct rate of the comprehensive testing reaching 95.6%. According to the national standard, the method of grade testing on peanut kernels´ specification and quality is designed, with which 100 grains of peanut are testing resulting with the result of the correct rate of the comprehensive testing reaching 92%. Using the method related in this article to test the appearance quality and distinguish the grade of specification can reach high correct rate which must produce positive significance to the peanut´s production and the industry´s development.