中国农业科学 ›› 2007, Vol. 40 ›› Issue (4): 698-703 .

• 农业信息技术 • 上一篇    下一篇

基于图像识别的玉米叶部病害诊断研究

赵玉霞,王克如,白中英,李少昆,谢瑞芝,高世菊   

  1. 中国农业科学院作物科学研究所/国家农作物基因资源与基因改良重大科学工程
  • 收稿日期:2006-05-22 修回日期:1900-01-01 出版日期:2007-04-10 发布日期:2007-04-10
  • 通讯作者: 李少昆

Maize Disease Identifying System Based Image Recognition

  

  1. 中国农业科学院作物科学研究所/国家农作物基因资源与基因改良重大科学工程
  • Received:2006-05-22 Revised:1900-01-01 Online:2007-04-10 Published:2007-04-10

摘要: 【目的】探讨利用图像技术实现玉米叶部病害自动识别的方法。【方法】根据玉米叶部病害特点,综合应用阈值法、区域标记方法与Freeman链码法,对玉米叶部病害图片进行图像分割、统计病斑个数、去除冗余斑点、计算病斑形状特征,最后根据二叉检索法推断病害。【结果】研究提取了五种玉米叶部主要病斑的识别特征,确定了诊断流程,并开发了识别系统。经检验,该系统对玉米叶部的锈病斑、弯孢菌病斑、灰斑、褐斑、小斑等五种主要病害的诊断准确率达80%以上。【结论】研究结果表明,用图像技术进行玉米叶部病害诊断是可行的,本研究开发的诊断系统为玉米病害自动识别与诊断奠定了基础。

关键词: 玉米叶部病害, 颜色特征, 形状特征, 图像分割, 区域标记

Abstract: 【Objective】 The recognition and diagnosis of main maize leaf disease is introduced by the technique of image recognition in the paper. 【Method】 It adopts threshold method to do image segmentation, and uses area-marking method calculating the num of disease as well as wiping off redundancy dots. And then it uses Freeman link code to calculate form feather. 【Result】The research presents the exclusive feather of main maize leaf disease and confirms the flow of disease diagnosis. The results indicate that the precision of six kinds of maize disease recognition is higher than 80%. 【Conclusion】 It shows that this method is available for recognizing maize disease. And it also provides technique support for the automatic recognition of maize disease by compiling the system with reasonable process flow.

Key words: Maize leaf disease, Color feather, Form feather, Image segmentation, Area marking