基于SVM的水稻稻瘟病识别方法  

Method of recognition of rice blast based on SVM

作　　者：赵开才[1] 石凤梅[2] 孟庆林[2] 马立功[2] 

机构地区：[1]黑龙江省科学技术情报研究所,哈尔滨150001 [2]黑龙江省农业科学院植物保护研究所,哈尔滨150080

出　　处：《东北农业大学学报》2013年第11期118-126,F0003,共10页Journal of Northeast Agricultural University

基　　金：黑龙江省青年基金项目(QC2009C69)

摘　　要：针对水稻稻瘟病人工识别准确性和效率不高的问题,提出基于多分类支持向量机的水稻稻瘟病识别方法。首先进行不同水稻稻瘟病病斑的颜色特征和形状特征提取,经过特征选择确定8个最佳特征组合,然后利用多分类支持向量机,对不同类型水稻稻瘟病进行识别。通过比较多分类支持向量机不同参数下的识别效果,确定稻瘟病识别支持向量机最佳模型参数。试验结果表明,基于多分类支持向量机的水稻稻瘟病识别方法具有较高识别精度,平均正确识别率达到了93.3%,能够有效地对水稻稻瘟病病害图像进行识别。A novel method for recognizing rice blast by using Multi-class support vector machines （multi-class SVM） was studied to improve recognition accuracy and efficiency. At first, the rice blast color features and shape features were extracted and the best eight feature sets were selected by analyzing. Different kind＇s of rice blast were classified based on the multi-class SVM, by using the best combined feature sets. The recognition accuracy using different classification kernel parameters were compared and analyzed, and the best model parameters were obtained. The experimental result show s that the rice blast recognition algorithm has obtained high recognizable accuracy based on multi-class SVM, the average correct recognition rate is 93.3%, can recognize the disease with different rice blast image.

关 键 词：图像识别 特征提取 多分类支持向量机 稻瘟病 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]