基于图像处理的温室黄瓜霜霉病诊断系统

发布时间：2018-05-04 13:18

本文选题：温室黄瓜 + 霜霉病　；参考：《农业机械学报》2017年02期

【摘要】：为进一步提高温室黄瓜霜霉病诊断的准确率,构建了一个基于图像处理的温室黄瓜霜霉病诊断系统。针对温室黄瓜栽培现场采集的病害图像,采用基于条件随机场(Conditional random fields,CRF)的图像分割方法进行病斑图像分割,并采用决策树模型扩展一元势函数,提高病斑图像分割的准确性;将分割后的病斑图像转换到HSV颜色空间并提取其颜色、纹理和形状等25个特征,利用粗糙集方法进行特征选择与优化;构建了基于径向基核函数的SVM分类器,准确地识别与诊断温室黄瓜霜霉病。系统试验验证结果表明,该系统采用的病斑分割方法,能够克服复杂背景和光照条件的影响,准确地提取病斑图像;采用粗糙集方法能够有效地选择分类特征,将25个初始特征减少到12个,提高了运行效率;黄瓜霜霉病识别准确率达到90%,能够满足设施蔬菜叶部病害诊断的需求。
[Abstract]:In order to improve the accuracy of greenhouse cucumber downy mildew diagnosis, a greenhouse cucumber downy mildew diagnosis system based on image processing was constructed. Based on conditional random field CRF (conditional random field CRF), the disease image was segmented in greenhouse cucumber field, and the decision tree model was used to expand the univariate potential function to improve the accuracy of disease spot image segmentation. The segmented image is transformed into HSV color space and 25 features such as color, texture and shape are extracted, and feature selection and optimization are carried out using rough set method. A SVM classifier based on radial basis function (RBF) kernel function is constructed. Accurate identification and diagnosis of cucumber downy mildew in greenhouse. The experimental results show that the segmentation method can overcome the influence of complex background and illumination conditions and extract the disease spot image accurately, and the classification feature can be selected effectively by rough set method. The operation efficiency was improved by reducing 25 initial characteristics to 12, and the accuracy of cucumber downy mildew identification reached 90%, which could meet the requirement of vegetable leaf disease diagnosis.
【作者单位】：中国农业大学信息与电气工程学院;食品质量与安全北京实验室;天津市农业科学院信息研究所;
【分类号】：S436.36;;TN941.1

【相似文献】