城郊地区面向对象的多传感器影像分类

Object Oriented Multi-Sensor Image Classification for Peri-Urban Areas

城郊地区地物类型较为复杂,利用单一传感器图像进行分类精度有限,而多传感器数据结合起来则可以"优势互补",有利于遥感影像的解译和分析。本文以武汉城郊某地块为实验对象,结合ALOS卫星三种传感器数据(包括全色、多光谱以及SAR影像)的信息,采用向对象的方法进行分类,以获取更高的分类精度。实验结果表明适当运用sar影像与光学影像结合起来进行分类,可以较好地区分出水体与阴影、水体与农作物等,对于道路、居民地等地物类型的分类也有所帮助。

Due to the more complex feature type in peri-urban areas,the classification accuracy by use of the single sensor image is limited and multi-sensor data can be used as the complementation in favor of the interpretation and analysis of remote sensing images.By taking Wuhan suburban plots as experimental subjects and combing three sensor data of ALOS satellite(including panchromatic,multispectral and SAR images),object classification is adopted in order to obtain higher classification accuracy.The experimental results show that the appropriate use of sar and optical image classification can better distinguish water body and shadow,water and crops,roads,residential areas and other surface feature,etc.