摘要
综合利用计算机视觉技术和BP神经网络技术,实现对粮仓害虫的无损检测。通过对粮仓害虫图像的CCD图像预处理,获取了几何特征和不变矩等15个特征参数,并通过优化选取其中七个参数输入神经网络进行训练。仿真结果表明训练网络对粮仓四类常见害虫的识别率达到了85%,得到了较好的识别结果。
With the full use of computer vision and neural networks, this paper presents an automatic classification method in the stored-grain pests. Through the pretreatment of the CCD image of stored-grain pests, we extracted eight shape features and seven changeless moments, and get seven features as the input parameters of neural networks by the feature selection. By use of this BP neural network model, an experiment for recognizing twenty samples of four kinds of stored-grain pests was performed, and the accurate recognition ratio was 85 %.
出处
《粮食储藏》
2007年第6期7-9,共3页
Grain Storage
基金
华中农业大学2004年度科技创新基金资助项目计划(项目编号:52204-04077)
关键词
粮仓害虫
特征提取
时间网络
分类
stored- grain pests, feature selection, BP neural network, classification
