湿地是生态系统的重要组成部分,及时、准确地获得湿地基础信息,对湿地的动态监测、保护与可持续利用及其它领域的研究具有重要意义。以三江平原东北部沼泽湿地为例,利用分类回归树(Classification and Regression Tree,CART)算法从训练...湿地是生态系统的重要组成部分,及时、准确地获得湿地基础信息,对湿地的动态监测、保护与可持续利用及其它领域的研究具有重要意义。以三江平原东北部沼泽湿地为例,利用分类回归树(Classification and Regression Tree,CART)算法从训练样本数据集中挖掘分类规则,集成遥感影像的光谱特征、纹理特征和地学辅助数据建立研究区湿地信息提取的决策树模型。用实测的GPS样本点对分类结果进行精度验证,并与最大似然监督分类方法(Maximum Likelihood Classi-fication,MLC)进行对比。结果表明,基于CART的决策树分类结果的总精度和Kappa系数分别为82.65%和0.7935,分类精度较MLC监督分类方法有明显提高,是内陆淡水沼泽湿地信息提取的有效手段。展开更多
Contourlet is a new 2-D signal analysis tool as well as wavelet analysis.It provides a flexible multi-resolution local and directional image expansion,and is more suitable for image processing than other math tools.In...Contourlet is a new 2-D signal analysis tool as well as wavelet analysis.It provides a flexible multi-resolution local and directional image expansion,and is more suitable for image processing than other math tools.In this paper,an IKONOS image with forests is decomposed by Contourlet transformation,then a method for extracting the boundary of forests from high-resolution remote sensing imagery is proposed associated watershed transformation and texture analysis.The experiment showed that the method could extract the boundary of forests accurately.展开更多
文摘Contourlet is a new 2-D signal analysis tool as well as wavelet analysis.It provides a flexible multi-resolution local and directional image expansion,and is more suitable for image processing than other math tools.In this paper,an IKONOS image with forests is decomposed by Contourlet transformation,then a method for extracting the boundary of forests from high-resolution remote sensing imagery is proposed associated watershed transformation and texture analysis.The experiment showed that the method could extract the boundary of forests accurately.