›› 2014, Vol. 27 ›› Issue (3): 48-.

• 论文 • 上一篇    下一篇

基于神经网络的枣树红蜘蛛识别研究

邱道尹,李俊霞,杨利涛   

  1. (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)
  • 出版日期:2014-03-15 发布日期:2014-03-11
  • 作者简介:邱道尹(1961—),男,教授,博士研究生,硕士生导师。研究方向:智能化信息系统。E-mail:263873563@qq.com。李俊霞(1987—),女,硕士研究生。研究方向:模式识别与智能系统。杨利涛(1989—),男,硕士研究生。研究方向:数字图像处理,模式识别。
  • 基金资助:

    郑州市科技计划基金资助项目(112PNYGC046)

Research on Recognition of Red Spider Based on Neural Network

QIU Daoyin1,LI Junxia1,YANG Litao2   

  1. (1.College of Electric Power,North China University of Water Conservancy and Hydroelectric Power,Zhengzhou 450011,China;2.College of Information Engineering,North China University of Water Resources and Electric Power,Zhengzhou 450011,China)
  • Online:2014-03-15 Published:2014-03-11

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。

关键词: 害虫识别, 图像处理, 神经网络, Matlab

Abstract:

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.

Key words: pest identification;image processing;neural network;Matlab

中图分类号: 

  • TP391