A knowledge-based system for classifying wetland types using ENVISAT ASAR and LANDSAT TM data

摘要

This paper presents a knowledge-based system (KBS) for classifying wetlands in the Pearl River Estuary using multi-polarized ENVISAT SAR and Landsat TM data. Remote sensing can be used to obtain wetland information, such as wetland types. However, there are serious confusions in wetland classification using conventional methods because some wetland types have similar spectral behavior. This paper develops a KBS method for integrating ENVISAT SAR with Landsat TM data to solve this problem. First, this KBS method has advantages over conventional supervised and unsupervised classification methods by conveniently incorporating expert knowledge and other ancillary GIS data. Secondly, the use of multi-polarized ENVISAT SAR can provide much richer information to improve the classification accuracy. The experiment has demonstrated that this approach can produce much better wetland classification results than the convention methods.