Hyperspectral Recognition of Rice Damaged by Rice Leaf Roller Based on Support Vector Machine

Abstract

Abstract: The spectra of healthy leaves and leaves damaged by the rice leaf roller were measured and analyzed by the method of continuum removal. In the range of 430-530 nm and 560-730 nm, the band depth and slope were extracted. Then the extracted parameters were chosen as the input vector of the support vector machine (SVM) to design a support vector classifier for the recognition of the leaves damaged by the rice leaf roller. The results confirmed that the classification precision of the SVM with radial basis function(RBF) kernel function was as high as 100% when γ and C were 0.25 and 1, respectively. This could provide theoretic basis for farmers to recognize the rice leaf damaged by the rice leaf roller ontime and control it effectively.

Key words: support vector machine, rice leaf roller, hyperspectral remote sensing, continuum removal, rice, insect damage

摘要： 对健康水稻叶片以及受稻纵卷叶螟危害后的水稻叶片进行了室内光谱的测定及分析。对430~530 nm和560~730 nm波段采用连续统去除的方法，分别提取了波深、斜率参量作为径向基核函数支持向量机的输入变量，利用LIBSVM软件包构建叶片高光谱识别模型。当参数γ和惩罚系数C分别取0.25和1时构建的径向基支持向量机模型的分类性能最佳，识别精度达100%。研究结果为实时水稻病虫害的早期监测以及田间管理提供了一定的理论基础。

关键词: 支持向量机, 稻纵卷叶螟, 高光谱遥感, 连续统去除, 水稻, 虫害

SHI Jingjing,LIU Zhanyu,ZHANG Lili,ZHOU Wan,HUANG Jingfeng. Hyperspectral Recognition of Rice Damaged by Rice Leaf Roller Based on Support Vector Machine [J]. Chinese Journal of Rice Science.

石晶晶, 刘占宇张莉丽, 周湾, 黄敬峰 . 基于支持向量机（SVM）的稻纵卷叶螟危害水稻高光谱遥感识别[J]. 中国水稻科学.

References

[1]苏祖芳, 周纪平,丁海红. 稻作诊断. 上海: 上海科学技术出版社, 2007: 211-213.

[2]刘宇, 王建强, 冯晓东, 等. 2007年全国稻纵卷叶螟发生实况分析与2008年发生趋势预测. 中国植保导刊, 2008, 28(7): 33-35.

[3]陈鹏程, 张建华, 雷勇辉. 高光谱遥感监测农作物病虫害研究进展. 农业信息科学, 2006, 20(2): 388389.

[4]浦瑞良, 宫鹏. 高光谱遥感及其应用. 北京: 高等教育出版社, 2000: 3-93.

[5]邢素丽, 张广录. 我国农业遥感的应用现状与展望. 农业工程学报, 2003, 19(6): 174-176.

[6]Kobayashi T, Kanda E, Kitada K, et al. Detection of rice panicle blast with multispectral radiometer and the potential of using airborne multispectral scanners. Phytopathology, 2001, 91: 316-323.

[7]吴曙雯, 王人潮, 陈晓斌, 等. 稻叶瘟对水稻光谱特性的影响研究. 上海交通大学学报：农业科学版, 2002, 20(1): 73-76.

[8]刘占宇, 孙华生, 黄敬峰. 基于学习矢量量化神经网络的水稻白穗和正常穗的高光谱识别. 中国水稻科学, 2007, 21(6): 658-662.

[9]Vapnik V N. 统计学习理论的本质. 张学工, 译. 北京: 清华大学出版社, 2000: 95-96.

[10]王海光, 马占鸿, 王韬, 等. 高光谱在小麦条锈病严重度分级识别中的应用.光谱学与光谱学分析, 2007, 27(9): 1811-1814.

[11]张录达, 苏时光, 王来生, 等. 支持向量机(SVM)在傅里叶变换近红外光谱分析中的应用研究. 光谱学与光谱学分析, 2005, 25(1): 33-35.

[12]Clark R N, King T V V, Ager C, et al. Initial vegetation species and senescence/stress indicator mapping in the San Luis Valley, Colorado using imaging spectrometer data//Posey H H, Pendelton J A, van Zyl D. Proceedings of the Summitville Forum ′95. Colorado Geological Survey Special Publication 38. Colorado: Colorado Geological Survey, Department of Natural Resource, 1995: 64-69.

[13]黄木易, 黄文江, 刘良云, 等. 冬小麦条锈病单叶光谱特性及严重度反演. 农业工程学报, 2004, 20(1): 176-180.

[14]张耀波, 张迁. 基于SVM的遥感影像的分类. 地理空间信息, 2005, 3(4): 24-26.

[15]田庆久, 宫鹏, 赵春江, 等. 用光谱反射率诊断小麦水分状况的可行性分析. 科技通报, 2000, 45(24): 2645-2650.

[16]Blakeman R H. The identification of crop disease and stress by aerial photography//Steven M D, Clark J A. Applications of Remote Sensing in Agriculture. Landon: Butterworths, 1990: 229245.

[17]Barnes J D, CardosaVilhena J. Interactions between electromagnetic radiation and the plant cuticle//Kerstiens G. Plant Cuticles: An integrated functional approach. Oxford: BIOS Scientific Publishers, 1996: 157170.

[18]Chang C C, Lin C J. LIBSVM: A library for support vector machines[CP/OL]. [2008-08-01]. http://www.csie.ntu.edu.tw/~cjlin/libsvm.