Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (13): 154-158.

Previous Articles     Next Articles

Automatic identification system for diseases of crop leaves based on image processing

LIU Jun1, WANG Zhenzhong1, LI Baoju2, HUAN Zhongdan1, HUANG Haiyang1   

  1. 1.Laboratory of Mathematics and Complex Systems, Ministry of Education, School of Mathematical Sciences, Beijing Normal University, Beijing 100875, China
    2.Institute of Vegetables and Flowers, Chinese Academy of Agriculture Sciences, Beijing 100081, China
  • Online:2012-05-01 Published:2012-05-09

基于图像处理的作物病害自动识别系统的研究

刘  君1,王振中1,李宝聚2,郇中丹1,黄海洋1   

  1. 1.北京师范大学 数学科学学院 数学与复杂系统教育部重点实验室,北京 100875
    2.中国农业科学院 蔬菜花卉研究所,北京 100875

Abstract: An automatic identification of the crop diseases system which is based on the images of infected leaves is proposed for the purpose of achieving the automatic control and detection of the crop diseases. This method is developed by extracting the whole and true lesions with image segmentation using the Expectation Maximization(EM) algorithm and Partial Differential Equation(PDE)-based level set method, and then some features of the lesions such as colors, shapes, textures are choosen to be studied and classified with the Principal Component Analysis(PCA), Neural Network(NN) and Support Vector Machine(SVM) methods. With these features and data analysis, different diseases can be automatically identified by the computer. This system has been tested and applied in some vegetables such as cucumber and tomato for detecting and identifying various diseases, and its high automatic level and recognition rate under some given conditions have been proved. 

Key words: diseases of leaves, automatic identification, image segmentation, EM algorithm, level set method

摘要: 为了实现对作物病害检测与防治的自动化,构建了一个基于叶片病斑图像处理的计算机诊断系统,以实现作物叶部病害的自动识别。该系统依据作物病叶颜色差异,用EM算法和偏微分方程水平集模型等图像分割算法,从图像中获取完整准确的病斑;然后提取病斑的颜色、形状和纹理特征,运用主成分分析方法对数据进行降维处理;最后采用神经网络和支持向量机方法对这些特征进行学习与分类,以及病害识别。系统已试用于黄瓜、番茄等园艺作物叶部病害的自动诊断与识别,其优点是自动化程度高,识别准确率在一定条件下较好。

关键词: 叶部病害, 自动识别, 图像分割, EM算法, 水平集方法