南极地区Landsat8卫星影像的阴影检测方法

Shadow detection in Antarctica using Landsat8 satellite images

阴影在光学遥感影像中普遍存在,会对后续一系列图像信息提取等过程产生影响.在南极地区,由于其特殊的地理位置和环境,光学遥感影像中存在的阴影不可忽视.本研究在已有的阴影检测方法的基础上,根据Landsat8卫星数据的特点构建了一个新的阴影检测方法:首先对所有数据进行反射率转换处理,利用南极地区成像时的阴影在各波段的反射率特征,选取第1波段Coastal/Aerosol(0.43~0.45μm)和第5波段NIR(0.85~0.88μm)建立了能够有效识别阴影的阴影检测指数R_(51),并选取不同地物的训练样本进行统计分析,根据阈值提取法提取阴影为二值化图像,并从影像中剔除少量易与阴影混淆的冰面融池.通过选取南极不同地区的影像进行精度评价,发现该方法针对Landsat8数据在南极地区的阴影检测具有较高的提取精度和较好的普适性.本文利用阴影检测方法对收集的覆盖南极地区的1 127景Landsat8影像进行阴影提取与统计、分布特征分析,最终提取阴影面积约15 000km^2,主要分布于裸岩山体集中区和大型冰川所在地.

Shadows present in optical remote sensing images will influence land surface information extraction and analysis. Such shadows are the result of ups and downs in the terrain and cannot be ignored especially in Antarctica located in high latitude with rather low solar elevation. Referring to existing shadow detection studies, the reflectance characteristics of shadows in Landsat8 images were analyzed for building a new method. DN values were converted to Top of Atmosphere （TOA） reflectance, and shadow detection index R51 were defined using handl Coastal/Aerosol（0. 43 - 0. 45 μm） and band5 NIR（0. 85 - 0. 88 μm） ; the threshold due to statistical analysis of training samples was used for differentiating shadow pixels with snow, blue ice and exposed rocks. Several images in different regions of Antarctica were used for accuracy assessment and the results suggested this method could be used for shadow detection of Landsat8 images in Antarctica with high accuracy. In addition, this method were applied for 1 127 Landsat8 images of the whole Antarctica continent, and the extraction shadow areas were nearly 15 000 kmz , most of which existed in rock areas and large glaciers regions.