小波域马铃薯典型虫害图像特征选择与识别

发布时间：2018-03-31 07:37

本文选题：马铃薯虫害　切入点：小波域　出处：《农业机械学报》2017年09期

【摘要】：为准确、快速地识别马铃薯典型虫害,提出了一种基于小波域的马铃薯典型虫害特征提取与识别方法。该方法以自然环境下的马铃薯虫害分割图像为对象,提取小波域高斯空间模型的高频协方差阵特征值与低频低阶矩(HELM)的12个不变纹理特征、空间域Hu不变矩的4个形状特征,进行支持向量机(SVM)的虫害分类识别。通过对8类典型虫害的识别,试验结果表明:在SVM识别方法下,本文HELM特征提取方法,相比传统纹理特征提取方法,在特征计算量不增加的同时,平均识别率至少提高了17个百分点;在HELM特征与Hu矩特征下,本文SVM的运行时间为0.481 s,比人工神经网络快了近2 s,平均识别率为97.5%,比人工神经网络、贝叶斯分类器识别率提高了至少6个百分点,有明显的识别优势。
[Abstract]:In order to accurately and quickly identify typical potato pests, a wavelet domain based feature extraction and recognition method for potato pests is proposed, which takes the image segmentation of potato pests in natural environment as an object. Extracting the eigenvalues of high frequency covariance matrix of Gao Si spatial model in wavelet domain and 12 invariant texture features of low frequency and low order moments, and 4 shape features of Hu invariant moments in spatial domain. Through the recognition of 8 kinds of typical insect pests, the experimental results show that under the SVM recognition method, the HELM feature extraction method in this paper, compared with the traditional texture feature extraction method, is better than the traditional texture feature extraction method. At the same time, the average recognition rate increased by at least 17 percentage points, and the average recognition rate increased by at least 17 percentage points under the HELM feature and Hu moment feature. The operating time of SVM is 0.481 s, which is nearly 2 s faster than that of artificial neural network, and the average recognition rate is 97.5%. The recognition rate of Bayesian classifier is 6 percentage points higher than that of artificial neural network and Bayesian classifier.
【作者单位】：内蒙古工业大学电力学院;
【基金】：国家自然科学基金项目(61661042) 内蒙古自治区自然科学基金项目(2015MS0617)
【分类号】：S435.32;TP391.41

【相似文献】