基于图像处理技术的四种苜蓿叶部病害的识别

发布时间：2018-01-30 02:47

本文关键词： 苜蓿叶部病害图像识别图像分割特征优选支持向量机　出处：《中国农业大学学报》2016年10期 　论文类型：期刊论文

【摘要】：基于图像处理技术,对4种苜蓿叶部病害进行识别研究。利用结合K中值聚类算法和线性判别分析的分割方法对病斑图像作分割,获得了较好的分割效果。结果表明:该分割方法在由4种病害图像数据集整合成的汇总图像数据集上综合得分的平均值和中值分别为0.877 1和0.899 7;召回率的平均值和中值分别为0.829 4和0.851 4;准确率的平均值和中值分别为0.924 9和0.942 4。进一步提取病斑图像的颜色特征、形状特征和纹理特征共计129个,利用朴素贝叶斯方法和线性判别分析方法建立病害识别模型,并结合顺序前向选择方法实现特征筛选,分别获得最优特征子集;同时利用这2个最优特征子集,结合支持向量机(Support vector machine,SVM)建立病害识别模型。比较各模型的识别效果,发现利用所建线性判别分析模型下的最优特征子集,结合SVM建立的病害识别模型识别效果最好,训练集识别正确率为96.18%,测试集识别正确率为93.10%。由此可见,本研究所建基于图像处理技术的病害识别模型可用于识别上述4种苜蓿叶部病害,为苜蓿病害的诊断和鉴别提供了一定依据。
[Abstract]:Based on image processing technology to study identification of 4 kinds of alfalfa leaf diseases. For the segmentation of the lesion image segmentation using median method combined with K clustering algorithm and linear discriminant analysis, obtain good segmentation results. The results show that the segmentation method in 4 kinds of diseases by image data integration summary image data sets the mean and median scores were 0.8771 and 0.8997; the mean and median recall rate were 0.8294 and 0.8514; the mean and median accuracy rate were 0.9249 and 0.942, 4. further lesion image color feature extraction, texture features and shape features a total of 129, using Naive Bayesian method and linear discriminant analysis method of disease recognition model, and combined with sequential forward selection method for feature selection, optimal feature subset is obtained respectively; at the same time using the 2 best feature subset, node Combined support vector machine (Support vector machine, SVM) to establish disease recognition model. The model recognition results, found that the use of the linear discriminant analysis model of the optimal feature subset, combined with disease identification model to identify the effect of SVM the best training set recognition correct rate is 96.18%, the correct recognition rate for the test set 93.10%. therefore, this study built the disease recognition model of image processing technology can be used to identify the 4 kinds of Alfalfa Leaf Diseases Based on, provides a basis for the diagnosis and differential diagnosis of diseases of alfalfa.

【作者单位】：中国农业大学植物保护学院;河北北方学院农林科技学院;中国科学院微生物研究所;
【基金】：公益性行业(农业)科研专项经费项目(201303057)
【分类号】：S435.4;S126
【正文快照】： 苜蓿被称为“牧草之王”,是我国最重要的栽培牧草,对畜牧业有重要价值[1,2]。苜蓿褐斑病(病原Pseudopeziza medicaginis)、锈病(病原Uromycesstriatus)、小光壳叶斑病(病原Pleosphaerulinabriosiana)和尾孢菌叶斑病(病原Cercosporamedicaginis)是4种常见的苜蓿叶部病害,影响植

【相似文献】