REMOTE SENSING CLASSIFICATION OF SURFACE FEATURES ON THE ROIG(E) PLATEAU
-
摘要: 各种物质在遥感数据中都有其特征的波谱吸收峰。因此通过波谱识别就可以大大减少以往遥感分类方法中的漏分及混分现象。为了解决若尔盖高原地物遥感分类问题,首次对本区ETM多光谱融合数据采用了地物波谱角分类法进行地物识别。分类步骤:遥感数据定标→选择训练区样本→MNF变换减维去噪处理→利用纯净像元指数(PPI)进行样本提纯→利用n-D散度法进行样本重组→波谱角分类(SAM).其分类结果表明,类别精度评价总精度为0.806,kappa系数为0.785.其精度均符合1:5万比例尺的土地详查工作的技术要求。Abstract: All materials have their characteristic spectral absorption peaks in remote sensing data.So missing and mixing of information that usually occurred in previous remote sensing classification may be greatly reduced by spectral identification.In order to solve the problem of the remote sensing classification of surface features on the Roigü Plateau, the spectral angle mapper classification of surface features was for the first time used in the ETM multispectral pansharp images.The steps of the classification are as follows:setting the standard of remote sensing data →selecting the swatches in the training area →using the minimum noise fraction rotation (MNF) to reduce the dimension and remove noises →using the pixel purity index (PPI) to purify the swatches →using the n-D visualizer to regroup the swatches →using the spectral angle mapper (ASM) to identify surface features.The classification results are as follows:the total accuracy of the class accuracy evaluation is 0.806 and the Kappa coefficient is 0.785.These accuracies conform to the technical requirements of 1:50 000 detailed land survey.
计量
- 文章访问数: 144
- HTML全文浏览量: 34
- PDF下载量: 9
- 被引次数: 0