多源遥感数据在植被覆盖区的水体信息提取

针对单一遥感影像难以准确提取南方植被覆盖区水体信息的问题,提出了先对光学影像进行混合像元分解,消除植被干扰,增强水体信息,再联合雷达影像的多源遥感数据提取方法。以地物类型丰富,水体发育的湖南衡阳地区为研究对象,选择Landsat8光学影像和PALSAR雷达影像为主要数据源,先利用混合像元分解模型从Landsat8光学影像提取出植被信息,对水体光谱进行重建,再综合去除植被的Landsat8影像的光谱信息和PALSAR影像的纹理信息进行面向对象的多源遥感水体信息提取,并与单源影像的水体信息提取结果进行对比。结果表明,该方法进行水体信息提取的精度更高,尤其在水陆过渡带,能够较好地去除植被干扰,提取被覆盖的水体信息,在南方植被覆盖区水体信息定量提取上具有很好的效果和潜力。

Water Body Information Extraction Based on Multisource Remote Sensing Data in Vegetation Coverage Area

In order to extract water information accurately from remote sensing imagery in southern vegetation coverage area, a new method has been applied to delineate and extract water body. Taking Hengyang as a case study, A linear model of spectral unmixing was applied in the remove of vegetation based on Landsat8 image. Then the Landsat8 image with vegetation removed and PALSAR image were fused to extract water body. Finally the accuracy of water body information derived from the fused multi-source remote sensing imagery was assessed using high-resolution imagery. Compared to single source water body information extraction results, the accuracy of the water information extraction method is higher than that of the single-source image, especially in the land-water transition zone, which can remove the vegetation disturbance and extract the covered water body information. The method has good effect and potential in the quantitative extraction of water body information in the vegetation cover area in the south.