基于ALOS遥感影像的湿地地表覆被信息提取研究

Study on Wetland Cover Information Extraction Based on ALOS Remote Sensing Image

为了验证ALOS遥感影像湿地地表覆被信息提取的可行性,以黑龙江省三江平原典型内陆淡水沼泽湿地为研究对象,通过ALOS遥感影像波段的光谱及纹理特性分析,探讨适合水体、旱地、水田、沼泽、林地、建设用地、草甸等覆被类型的分类特征;基于非监督、监督及面向对象分类方法,遴选能实现最优分类结果的特征组合,为湿地地表覆盖分类数据源及方法的选择提供参考。结果表明:非监督、监督及面向对象分类方法的总体精度分别达到63.86%、96.14%和85.26%;非监督分类方法整体分类效果不够理想;面向对象方法虽然得到了相对较高的分类精度,但是针对建设用地、林地及草甸地类信息提取的精度处于较低水平;监督分类方法能取得较好效果,最适合于湿地地表覆被信息提取。

In order to verify the feasibility of wetland cover information extraction for ALOS remote sensing image,classic inland freshwater wetland in Sanjiang Plain was taken as an example,spectral and textural characteristics of the image were analyzed,and classification characteristics of different cover types were discussed.Based on the methods of unsupervised,supervised and object-oriented classifications,the optimum combination of characteristics was selected to provide references for the data and method selections of wetland cover classification.The results showed that the accuracy with unsupervised,supervised and object-oriented classifications reached 63.86%,96.14% and 85.26%,respectively;characteristics were poor by means of unsupervised classification;the accuracy with object-oriented classification was very high,but construction,wood and grass cover types were differentiated difficultly;supervised classification was very suitable for wetland cover information extraction.