基于SVM的县域冬小麦种植面积遥感提取
|
摘要
冬小麦种植面积的精确提取,对于农业部门进行冬小麦生长监测与产量估测有着重要的支撑作用。本研究在对Landsat-8卫星15 m×15 m空间分辨率遥感影像进行预处理的基础上,基于最佳波段指数(OIF),采用支持向量机(SVM)算法中四种核函数进行影像分类,并比较分类精度,选择精度最高的核函数作为SVM最优核函数对盐城市大丰区冬小麦种植面积进行提取,与最大似然法、最小距离法的结果进行对比。结果表明,四种核函数中,Linear核函数分类精度最高,达到98.56%。将Linear核函数作为SVM最优核函数对大丰区冬小麦种植面积进行提取,提取到的种植面积为71 834.4 hm~2,提取精度、分类精度和Kappa系数分别为91.25%、98.56%和0.98。基于SVM的冬小麦面积提取效果明显好于传统监督分类方法,说明使用支持向量机与影像光谱特征进行影像分类能够准确提取县域冬小麦种植面积。
The accurate extraction of winter wheat area plays an important role in the winter wheat growth monitoring and yield estimation in the agricultural sector. In this study, the 15 m×15 m spatial resolution remote sensing image from Landsat-8 satellite was preprocessed. Then, based on the optimal band index(OIF), four kernel functions in support vector machine(SVM) algorithm were used in image classification.Their classification accuracies were compared. The highest precision kernel function was selected as the optimal kernel function of SVM to extract the winter wheat planting area in Dafeng District, and compared with the results of maximum likelihood method and minimum distance method. The results showed that among the four kernel functions, the Linear kernel function had the highest classification accuracy. And it was 98.56%. The winter wheat planting area extracted in Dafeng District with the Linear kernel function as the SVM optimal kernel function was 71 834.4 hm~2. The extraction precision accuracy, classification accuracy and Kappa coefficient were 91.25%, 98.56% and 0.98, respectively. The extraction effect of Winter wheat area based on SVM was significantly better than the traditional supervised classification method. It was suggested that the image classification based on support vector machine and image spectral features could accurately extract winter wheat planting area. The method could provide method support for the extraction of winter wheat area in the counties of Jianghuai region.
The accurate extraction of winter wheat area plays an important role in the winter wheat growth monitoring and yield estimation in the agricultural sector. In this study, the 15 m×15 m spatial resolution remote sensing image from Landsat-8 satellite was preprocessed. Then, based on the optimal band index(OIF), four kernel functions in support vector machine(SVM) algorithm were used in image classification.Their classification accuracies were compared. The highest precision kernel function was selected as the optimal kernel function of SVM to extract the winter wheat planting area in Dafeng District, and compared with the results of maximum likelihood method and minimum distance method. The results showed that among the four kernel functions, the Linear kernel function had the highest classification accuracy. And it was 98.56%. The winter wheat planting area extracted in Dafeng District with the Linear kernel function as the SVM optimal kernel function was 71 834.4 hm~2. The extraction precision accuracy, classification accuracy and Kappa coefficient were 91.25%, 98.56% and 0.98, respectively. The extraction effect of Winter wheat area based on SVM was significantly better than the traditional supervised classification method. It was suggested that the image classification based on support vector machine and image spectral features could accurately extract winter wheat planting area. The method could provide method support for the extraction of winter wheat area in the counties of Jianghuai region.
引文
[1]李卫国,赵丽花.中高分辨率遥感影像在小麦监测中的比较 [J].江苏农业学报,2011,27(4):736.LI W G,ZHAO L H.Wheat growth monitoring based on medium and high resolution images [J].Jiangsu Journal of Agriculture Sciences,2011,27(4):736.
[2]林子晶,李卫国,申双和,等.HJ星和GF1号数据在水稻种植面积提取中的应用 [J].江苏农业学报,2016,32(1):111.LIN Z J,LI W G,SHEN S H,et al.Application of HJ and GF1 image data to extract rice planting area [J].Jiangsu Journal of Agriculture Sciences,2016,32(1):111.
[3]王尔美,李卫国,顾晓鹤,等.基于光谱特征分异的玉米种植面积提取 [J].江苏农业学报,2017,33(4):822.WANG E M,LI W G,GU X H,et al.Planting area extraction of maize based on spectral features differentiation [J].Jiangsu Journal of Agriculture Sciences,2017,33(4):822.
[4]李晓东,姜琦刚.基于多时相遥感数据的农田分类提 [J].农业工程学报,2015,31(7):145.LI X D,JIANG Q G.Extraction farmland classification based on multi-temporal remote sensing data [J].Journal of Agricultural Engineering,2015,31(7):145.
[5]赵春霞,钱乐祥.遥感影像监督分类与非监督分类的比较 [J].河南大学学报,2004,34(3):90.ZHAO C X,QIAN L X.Comparative study of supervised and unsupervised classification in remote sensing image [J].Journal of Henan University(Natural Science),2004,34(3):90.
[6]王小明,毛梦琪,张昌景,等.基于支持向量机的遥感影像分类比较研究 [J].测绘与空间地理信息,2013,36(4):17.WANG X M,MAO M Q,ZHANG C J,et al.Comparative study on classification of remote sensing image by support vector machine [J].Geomatics Spatial Information Technology,2013,36(4):17.
[7]任琼.基于SVM的余杭生态公益林类型的遥感分类研究[D].南京:南京林业大学,2008:1.REN Q.Remote sensing classification of Yuhang ecological public welfare forest type based on SVM [D].Nanjing:Nanjing Forestry University,2008:1.
[8]FOODY G M,MATHUR A.The use of small training sets containing mixed pixels for accurate hard image classification:Training on mixed spectral responses for classification by a SVM [J].Remote Sensing of Environment,2006,103(2):179.
[9]董金芳,王娟,何慧娟,等.基于支持向量机的湿地遥感分类方法 [J].测绘与空间地理信息,2016,39(11):150.DONG J F,WANG J,HE H J,et al.Remote sensing classification for wetland based onsupport vector machine [J].Geomatics Spatial Information Technology,2016,39(11):150.
[10]李梦颖,邢艳秋,李美爽,等.基于支持向量机的Landsat-8影像森林类型识别研究 [J].中南林业科技大学学报,2017,37(4):52.LI M Y,XING Y Q,LIU M S,et al.Identification of forest type with Landsat-8 image based on SVM [J].Journal of Central South University of Forestry & Technology,2017,37(4):52.
[11]马鹏鹏,周爱明,姚青,等.图像特征和样本量对水稻害虫识别结果的影响 [J].中国水稻科学,2018,32(4):405.MA P P,ZHOU A M,YAO Q,et al.Influence of image features and sample sizes on rice pest identification [J].Chinese Journal of Rice Science,2018,32(4):405.
[12]李卫国,蒋楠.基于面向对象分类的冬小麦种植面积提取 [J].麦类作物学报,2012,32(4):701.LI W G,JIANG N.Extraction of winter wheat planting area by object-oriented classification method [J].Journal of Triticeae Crop,2012,32(4):701.
[13]YU T Z,XIANG M X.Mapping paddy rice planting area in rice-wetland coexistent areas through analysis of Landsat 8 OLI and MODIS [J].International Journal of Applied Earth Observation and Geinformation,2016,46:1.
[14]BING W Q,WEI J L.Mapping paddy rice areas based on vegetation phenology and surface moisture conditions [J].Ecological Indicators,2015,56:79.
[15]邓书斌.ENVI遥感图像处理方法[M].北京:高等教育出版社,2014:73.DENG S B.ENVI remote sensing image processing method[M].Beijing:Higher Education Press,2014:73.
[16]CHAVEZ P S,BERLIN G L,SOWERS L B.Statistical method for selecting Landsat MSS ratios [J].Journal of Applied Photographic Engineering,1982(8):23.
[17]VAPNIK V N.An overview of statistical learning theory [J].IEEE Transactions on Neural Networks,1999,10(5):988.
[18]WANG J H.New algorithm of remote sensing image classification based on K-type support vector support vector machine [J].Computer Applications,2012,32(10):2832.
[19]王振武,孙佳俊,于忠义,等.基于支持向量机的遥感图像分类研究综述 [J].计算机科学,2016,43(9):11.WANG Z W,SUN J J,YU Z Y,et al.Review of remote sensing image classification based on support vector machine [J].Computer Science,2016,43(9):11.
[20]李帅,张梦华,郭力娜.不同监督分类器对土地利用分类精度的影响 [J].华北理工大学学报,2018,40(2):42.LI S,ZHANG M H,GUO L N.Effect of different supervised classifiers on land use classification accuracy [J].Journal of North China University of Technology,2018,40(2):42.
[21]葛广秀,李卫国,景元书.基于NDVI密度分割的冬小麦种植面积提取 [J].麦类作物学报,2014,34(7):997.GE G X,LI W G,JING Y S.Rea of winter wheat extracted on ndvi density slicing [J].Journal of Triticeae Crops,2014,34(7):997.
[22]王利民,刘佳,高建孟,等.冬小麦面积遥感识别精度与空间分辨率的关系 [J].农业工程学报,2016,32(23):152.WANG L M,LIU J,GAO J M,et al.Remote sensing classification of marsh wetland with different resolution images [J].Journal of Agricultural Engineering,2016,32(23):152.
[2]林子晶,李卫国,申双和,等.HJ星和GF1号数据在水稻种植面积提取中的应用 [J].江苏农业学报,2016,32(1):111.LIN Z J,LI W G,SHEN S H,et al.Application of HJ and GF1 image data to extract rice planting area [J].Jiangsu Journal of Agriculture Sciences,2016,32(1):111.
[3]王尔美,李卫国,顾晓鹤,等.基于光谱特征分异的玉米种植面积提取 [J].江苏农业学报,2017,33(4):822.WANG E M,LI W G,GU X H,et al.Planting area extraction of maize based on spectral features differentiation [J].Jiangsu Journal of Agriculture Sciences,2017,33(4):822.
[4]李晓东,姜琦刚.基于多时相遥感数据的农田分类提 [J].农业工程学报,2015,31(7):145.LI X D,JIANG Q G.Extraction farmland classification based on multi-temporal remote sensing data [J].Journal of Agricultural Engineering,2015,31(7):145.
[5]赵春霞,钱乐祥.遥感影像监督分类与非监督分类的比较 [J].河南大学学报,2004,34(3):90.ZHAO C X,QIAN L X.Comparative study of supervised and unsupervised classification in remote sensing image [J].Journal of Henan University(Natural Science),2004,34(3):90.
[6]王小明,毛梦琪,张昌景,等.基于支持向量机的遥感影像分类比较研究 [J].测绘与空间地理信息,2013,36(4):17.WANG X M,MAO M Q,ZHANG C J,et al.Comparative study on classification of remote sensing image by support vector machine [J].Geomatics Spatial Information Technology,2013,36(4):17.
[7]任琼.基于SVM的余杭生态公益林类型的遥感分类研究[D].南京:南京林业大学,2008:1.REN Q.Remote sensing classification of Yuhang ecological public welfare forest type based on SVM [D].Nanjing:Nanjing Forestry University,2008:1.
[8]FOODY G M,MATHUR A.The use of small training sets containing mixed pixels for accurate hard image classification:Training on mixed spectral responses for classification by a SVM [J].Remote Sensing of Environment,2006,103(2):179.
[9]董金芳,王娟,何慧娟,等.基于支持向量机的湿地遥感分类方法 [J].测绘与空间地理信息,2016,39(11):150.DONG J F,WANG J,HE H J,et al.Remote sensing classification for wetland based onsupport vector machine [J].Geomatics Spatial Information Technology,2016,39(11):150.
[10]李梦颖,邢艳秋,李美爽,等.基于支持向量机的Landsat-8影像森林类型识别研究 [J].中南林业科技大学学报,2017,37(4):52.LI M Y,XING Y Q,LIU M S,et al.Identification of forest type with Landsat-8 image based on SVM [J].Journal of Central South University of Forestry & Technology,2017,37(4):52.
[11]马鹏鹏,周爱明,姚青,等.图像特征和样本量对水稻害虫识别结果的影响 [J].中国水稻科学,2018,32(4):405.MA P P,ZHOU A M,YAO Q,et al.Influence of image features and sample sizes on rice pest identification [J].Chinese Journal of Rice Science,2018,32(4):405.
[12]李卫国,蒋楠.基于面向对象分类的冬小麦种植面积提取 [J].麦类作物学报,2012,32(4):701.LI W G,JIANG N.Extraction of winter wheat planting area by object-oriented classification method [J].Journal of Triticeae Crop,2012,32(4):701.
[13]YU T Z,XIANG M X.Mapping paddy rice planting area in rice-wetland coexistent areas through analysis of Landsat 8 OLI and MODIS [J].International Journal of Applied Earth Observation and Geinformation,2016,46:1.
[14]BING W Q,WEI J L.Mapping paddy rice areas based on vegetation phenology and surface moisture conditions [J].Ecological Indicators,2015,56:79.
[15]邓书斌.ENVI遥感图像处理方法[M].北京:高等教育出版社,2014:73.DENG S B.ENVI remote sensing image processing method[M].Beijing:Higher Education Press,2014:73.
[16]CHAVEZ P S,BERLIN G L,SOWERS L B.Statistical method for selecting Landsat MSS ratios [J].Journal of Applied Photographic Engineering,1982(8):23.
[17]VAPNIK V N.An overview of statistical learning theory [J].IEEE Transactions on Neural Networks,1999,10(5):988.
[18]WANG J H.New algorithm of remote sensing image classification based on K-type support vector support vector machine [J].Computer Applications,2012,32(10):2832.
[19]王振武,孙佳俊,于忠义,等.基于支持向量机的遥感图像分类研究综述 [J].计算机科学,2016,43(9):11.WANG Z W,SUN J J,YU Z Y,et al.Review of remote sensing image classification based on support vector machine [J].Computer Science,2016,43(9):11.
[20]李帅,张梦华,郭力娜.不同监督分类器对土地利用分类精度的影响 [J].华北理工大学学报,2018,40(2):42.LI S,ZHANG M H,GUO L N.Effect of different supervised classifiers on land use classification accuracy [J].Journal of North China University of Technology,2018,40(2):42.
[21]葛广秀,李卫国,景元书.基于NDVI密度分割的冬小麦种植面积提取 [J].麦类作物学报,2014,34(7):997.GE G X,LI W G,JING Y S.Rea of winter wheat extracted on ndvi density slicing [J].Journal of Triticeae Crops,2014,34(7):997.
[22]王利民,刘佳,高建孟,等.冬小麦面积遥感识别精度与空间分辨率的关系 [J].农业工程学报,2016,32(23):152.WANG L M,LIU J,GAO J M,et al.Remote sensing classification of marsh wetland with different resolution images [J].Journal of Agricultural Engineering,2016,32(23):152.