基于神经网络的枣树红蜘蛛识别研究  

Research on Recognition of Red Spider Based on Neural Network

作　　者：邱道尹[1] 李俊霞[1] 杨利涛[2] 

机构地区：[1]华北水利水电大学电力学院,河南郑州450011 [2]华北水利水电大学信息工程学院,河南郑州450011

出　　处：《电子科技》2014年第3期48-51,共4页Electronic Science and Technology

基　　金：郑州市科技计划基金资助项目(112PNYGC046)

摘　　要：针对目前枣树病虫害防治过程中害虫的自动识别问题,文中以枣树害虫红蜘蛛为研究对象,介绍一种基于BP神经网络技术和图像处理进行枣树虫害识别的方法,即利用图像处理技术,结合枣虫害颜色特征、几何特征实现枣树虫害图像的特征提取。并利用Matlab完成对神经网络的训练和测试,构建图像自动分类器,实现枣树虫害图像的自动识别。同时通过对实际采集到的图像进行验证,识别率结果达到了95.3%,证明了文中所提方法的有效性和鲁棒性。This paper introduces a method for recognizing jujube pest based on technologies of BP neural network and digital image processing for the automatic recognition of pests in preventing the jujube pests. Using image processing, this paper accomplishes the feature extraction combining with color characteristics and geometrical characteristics of jujube pest image. Using Matlab, this paper fulfills the training and testing of the neural network, constructs the automatic image classification, and realizes the automatic identification of insect pest images. Experiments on actual images show that proposed method successfully recognizes the most parking lines with a detection rate of 95.3%, which proves its effectiveness and robustness.

关 键 词：害虫识别 图像处理 神经网络 MATLAB 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]