基于决策树判别的高温目标遥感识别方法
|
摘要
通常,高温目标与常温地物间光谱特征差异显著;但研究发现,Landsat8 OLI遥感影像中彩钢屋顶像元却与高温目标(林火)的光谱特征颇为相似,使用以往高温目标识别方法识别效果不佳。为实现高温目标的精确识别,引入决策树判别法;根据不同地物类型的相似程度构建决策树模型,针对各分支结点的相似地物类型,按定量指标分别进行特征波段筛选,确定反映地物间本质区别的判别函数,并经分类统计确定判别阈值。研究表明,所构建的决策树能够准确划分地物类型,在实现同一般常温地物有效区分的同时,能有针对性地区分高温目标与彩钢屋顶建筑,高温目标识别精度为97. 67%。
Generally,spectral differences between high temperature targets and normal temperature objects are significant. However,it is found that pixels of color steel roof in the Landsat8 OLI remote sensing image are quite similar to the high temperature targets( forest fires),and using previous high temperature target recognition method to identify not effectively. In order to realize accurate identification of high temperature targets,decision tree discriminant method is introduced. The decision tree model is constructed according to the similarity degree of different feature types. For the similar feature types under each branch node,sensitive bands are screened according to quantitative indicators,discriminant functions reflecting the essential difference between objects are determined,and discriminant thresholds are determined by classification statistics. The research shows that the constructed decision tree can accurately classify feature types,and classify high temperature targets and color steel roof buildings in a targeted way while achieving effective separation from normal temperature objects. The high temperature targets recognition accuracy is 97. 67%.
Generally,spectral differences between high temperature targets and normal temperature objects are significant. However,it is found that pixels of color steel roof in the Landsat8 OLI remote sensing image are quite similar to the high temperature targets( forest fires),and using previous high temperature target recognition method to identify not effectively. In order to realize accurate identification of high temperature targets,decision tree discriminant method is introduced. The decision tree model is constructed according to the similarity degree of different feature types. For the similar feature types under each branch node,sensitive bands are screened according to quantitative indicators,discriminant functions reflecting the essential difference between objects are determined,and discriminant thresholds are determined by classification statistics. The research shows that the constructed decision tree can accurately classify feature types,and classify high temperature targets and color steel roof buildings in a targeted way while achieving effective separation from normal temperature objects. The high temperature targets recognition accuracy is 97. 67%.
引文
1袁悦.高温目标短波红外遥感识别方法改进研究[D].长春:吉林大学,2015Yuan Yue.The study of high-temperature targets identification method improvement in shortwave infrared remote sensing[D].Changchun:Jilin University,2015
2袁悦,潘军,邢立新,等.基于Fisher两类判别的高温目标精确识别方法[J].科学技术与工程,2015,15(9):109-113Yuan Yue,Pan Jun,Xing Lixin,et al.Identification of high temperature targets based on Fisher two types discrimination[J].Science Technology and Engineering,2015,15(9):109-113
3朱亚静.高温地物目标短波红外遥感识别及温度反演[D].长春:吉林大学,2012Zhu Yajing.High-temperature target identification and temperature retrieval using shortwave infrared remote sensing data[D].Changchun:Jilin University,2012
4于一凡,潘军,邢立新,等.基于马氏距离的遥感图像高温目标识别方法研究[J].遥感信息,2013,28(5):90-94Yu Yifan,Pan Jun,Xing Lixin,et al.Identification of high temperature targets in remote sensing imagery based on Mahalanobis distance[J].Remote Sensing Information,2013,28(5):90-94
5于一凡,潘军,邢立新,等.短波红外波段高温目标识别的可行性分析[J].国土资源遥感,2014,26(1):25-30Yu Yifan,Pan Jun,Xing Lixin,et al.Feasibility analysis of shortwave infrared band for recognition of high temperature target[J].Remote Sensing for Land&Resources,2014,26(1):25-30
6于一凡.短波红外遥感高温目标温度反演模型研究[D].长春:吉林大学,2014Yu Yifan.The study of high-temperature targets temperature retrieval model in shortwave infrared remote sensing[D].Changchun:Jilin University,2014
7 Yu Y,Pan J,Xing L,et al.Identification of high temperature targets in remote sensing imagery based on factor analysis[J].Journal of Applied Remote Sensing,2014,8(1):1-8
8杨超,邬国锋,李清泉,等.植被遥感分类方法研究进展[J].地理与地理信息科学,2018,34(4):24-32Yang Chao,Wu Guofeng,Li Qingquan,et al.Research progress on remote sensing classification of vegetation[J].Geography and GeoInformation Science,2018,34(4):24-32
9 Hansen M,Dubayah R,De Fries R.Classification trees,an alternative to traditional land cover classifiers[J].International Journal of Remote Sensing,1996,17:1075-1081
10 Quinlan J R.C4.5:programs for machine learning[M].San Mateo.CA:Morgan Kaufman Publishers,1993
11李德仁,王树良,李德毅,等.论空间数据挖掘和知识发现的理论与方法[J].武汉大学学报信息科学版,2002,27(3):221-233Li Deren,Wang Shuliang,Li Deyi,et al.Theories and technologies of spatial data mining and knowledge discovery[J].Geomatics and Information Science of Wuhan University,2002,27(3):221-233
12 Duda R O,Hart P E,Stork D G.模式分类[M].李宏东,姚天翔,译.北京:机械工业出版社,2003Duda R O,Hart P E,Stork D G.Pattern classification[M].Li Hongdong,Yao Tianxiang,translated.Beijing:Mechanical Industry Press,2003
13 Friedl M A,Brodeley C E.Decision tree classification of land cover from remotely sensed data[J].Remote Sensing Environment,1997,61:399-409
14纪宏金,时艳香,陆继龙.地球化学数据统计分析[M].长春:吉林大学出版社,2014:138-145Ji Hongjin,Shi Yanxiang,Lu Jilong.The analysis of geochemical data statistics[M].Changchun:Jilin University Press,2014:138-145
15王鹏举,潘军,蒋立军,等.林火目标遥感多光谱识别指数构建方法研究[J].科学技术与工程,2018,18(2):312-319Wang Pengju,Pan Jun,Jiang Lijun,et al.Method of remote sensing multispectral recognition index construction for forest fire[J].Science Technology and Engineering,2018,18(2):312-319
2袁悦,潘军,邢立新,等.基于Fisher两类判别的高温目标精确识别方法[J].科学技术与工程,2015,15(9):109-113Yuan Yue,Pan Jun,Xing Lixin,et al.Identification of high temperature targets based on Fisher two types discrimination[J].Science Technology and Engineering,2015,15(9):109-113
3朱亚静.高温地物目标短波红外遥感识别及温度反演[D].长春:吉林大学,2012Zhu Yajing.High-temperature target identification and temperature retrieval using shortwave infrared remote sensing data[D].Changchun:Jilin University,2012
4于一凡,潘军,邢立新,等.基于马氏距离的遥感图像高温目标识别方法研究[J].遥感信息,2013,28(5):90-94Yu Yifan,Pan Jun,Xing Lixin,et al.Identification of high temperature targets in remote sensing imagery based on Mahalanobis distance[J].Remote Sensing Information,2013,28(5):90-94
5于一凡,潘军,邢立新,等.短波红外波段高温目标识别的可行性分析[J].国土资源遥感,2014,26(1):25-30Yu Yifan,Pan Jun,Xing Lixin,et al.Feasibility analysis of shortwave infrared band for recognition of high temperature target[J].Remote Sensing for Land&Resources,2014,26(1):25-30
6于一凡.短波红外遥感高温目标温度反演模型研究[D].长春:吉林大学,2014Yu Yifan.The study of high-temperature targets temperature retrieval model in shortwave infrared remote sensing[D].Changchun:Jilin University,2014
7 Yu Y,Pan J,Xing L,et al.Identification of high temperature targets in remote sensing imagery based on factor analysis[J].Journal of Applied Remote Sensing,2014,8(1):1-8
8杨超,邬国锋,李清泉,等.植被遥感分类方法研究进展[J].地理与地理信息科学,2018,34(4):24-32Yang Chao,Wu Guofeng,Li Qingquan,et al.Research progress on remote sensing classification of vegetation[J].Geography and GeoInformation Science,2018,34(4):24-32
9 Hansen M,Dubayah R,De Fries R.Classification trees,an alternative to traditional land cover classifiers[J].International Journal of Remote Sensing,1996,17:1075-1081
10 Quinlan J R.C4.5:programs for machine learning[M].San Mateo.CA:Morgan Kaufman Publishers,1993
11李德仁,王树良,李德毅,等.论空间数据挖掘和知识发现的理论与方法[J].武汉大学学报信息科学版,2002,27(3):221-233Li Deren,Wang Shuliang,Li Deyi,et al.Theories and technologies of spatial data mining and knowledge discovery[J].Geomatics and Information Science of Wuhan University,2002,27(3):221-233
12 Duda R O,Hart P E,Stork D G.模式分类[M].李宏东,姚天翔,译.北京:机械工业出版社,2003Duda R O,Hart P E,Stork D G.Pattern classification[M].Li Hongdong,Yao Tianxiang,translated.Beijing:Mechanical Industry Press,2003
13 Friedl M A,Brodeley C E.Decision tree classification of land cover from remotely sensed data[J].Remote Sensing Environment,1997,61:399-409
14纪宏金,时艳香,陆继龙.地球化学数据统计分析[M].长春:吉林大学出版社,2014:138-145Ji Hongjin,Shi Yanxiang,Lu Jilong.The analysis of geochemical data statistics[M].Changchun:Jilin University Press,2014:138-145
15王鹏举,潘军,蒋立军,等.林火目标遥感多光谱识别指数构建方法研究[J].科学技术与工程,2018,18(2):312-319Wang Pengju,Pan Jun,Jiang Lijun,et al.Method of remote sensing multispectral recognition index construction for forest fire[J].Science Technology and Engineering,2018,18(2):312-319