关键词: 图像处理, 颜色特征, 大豆, 粒子群优化算法, 机器视觉

Abstract: The new national standard of soybean(GB 1352-2009) is formally promulgated and implemented on September 1, 2009. Based on this fact that heat-damaged soybean is first proposed, a classification method by utilizing neural network and soybean surface color features is presented. This article chooses six color feature data to use as the input of the classifier and tries to use a new algorithm which combines particle swarm optimization with BP (Back Propagation) to train neural network. The simulation results show that the proposed method is of certain effect of soybean, which is helpful to realize the online vulnerability grain detection.

Key words: image processing, color properties, soybean, particle swarm optimization algorithm, machine vision