| 高分六号遥感数据在有林地识别中的应用 |
| |
| 引用本文： | 梁志国,隋傲,于颖,赵戈榕,谢秋,刘代超.高分六号遥感数据在有林地识别中的应用[J].东北林业大学学报,2020(5):35-39. |
| |
| 作者姓名： | 梁志国 隋傲 于颖 赵戈榕 谢秋 刘代超 |
| |
| 作者单位： | 东北林业大学林学院;东北林业大学;中国科学院空天信息创新研究院 |
| |
| 基金项目： | 国家高分辨率对地观测系统重大专项(21-Y20A06-9001-17/18);国家大学生创新训练项目(201810225489)。 |
| |
| 摘 要： | 精准农业观测卫星-高分六号卫星(GF6)增加了4个特殊波段,更加有效地反映了植被特有的光谱特性,为植被应用研究提供更为详细的地物光谱信息。为了分析GF6数据在植被识别能力上的优越性,比较了GF6号新增波段(红边1、红边2、黄边、紫边波段)和高分数据传统波段对有林地识别精度的影响。结果表明:GF6新增波段对有林地快速识别的精度达到97.67%,Kappa系数为0.95,比GF数据4波段对有林地的识别精度提高了3.35%,Kappa系数提高了0.08。CART自适应特征和阈值选择决策树算法比人工决策树分类算法对有林地识别精度有显著增加,精度由88.81%提高到97.67%,Kappa系数由0.78提高到0.95。GF6数据新增特殊波段结合CART自适应特征和阈值决策树算法对有林地具有快速优越的识别能力。
|
| 关 键 词： | 高分六号卫星 植被分类 决策树 决策树算法 |
Rapid Identification of Forestland with GF6 Data |
| |
| Liang Zhiguo,Sui Ao,Yu Ying,Zhao Gerong,Xie Qiu,Liu Daichao.Rapid Identification of Forestland with GF6 Data[J].Journal of Northeast Forestry University,2020(5):35-39. |
| |
| Authors: | Liang Zhiguo Sui Ao Yu Ying Zhao Gerong Xie Qiu Liu Daichao |
| |
| Institution: | (Northeast Forestry University,Harbin 150040,P.R.China;Aerospace Information Research Institute Chinese Academy of Sciences) |
| |
| Abstract: | Four special bands in Gaofen-6 satellite(GF6)were added for the first time,which reflected the unique spectral characteristics of vegetation more effectively and provided more detailed spectral information for vegetation application research.In order to highlight the superiority of GF6 data in vegetation identification ability,the effects of the newly added bands of GF6(red band 1 and 2,yellow band and purple band)and the traditional bands of GF6 data on the identification accuracy of woodland were compared.In order to further improve the accuracy of forest land identification,the difference between artificial decision tree classification and CART algorithm based decision tree classification was studied.The accuracy of the new band of GF6 for the rapid identification of forest land was 97.67%,and the Kappa coefficient was 0.95,which was 3.35%higher than that of the traditional four-band of GF data,and the Kappa coefficient was 0.08 higher.Compared with the artificial decision tree classification algorithm,CART adaptive feature and threshold selection decision tree algorithm can increase the recognition accuracy of forest land,with the accuracy increased from 88.81%to 97.67%,and Kappa coefficient increased from 0.78 to 0.95.It can be seen that the new special band in GF6 data,combined with CART adaptive features and threshold decision tree algorithm,has a fast and superior identification ability for forestland. |
| |
| Keywords: | GF-6 satellite New bands Decision tree classification CART Algorithm |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
