微光卫星遥感图像光度校正技术改进研究

An improvement on the illumination correction method for low-light imagery is proposed in this application and this new method aims to improve low-light imagery product quality for domestic satellite. In recent years, low-light satellite imagery systems, such as OLS on DMSP satellites and VIIRS on S-NPP satellite, attract world-wide attentions. These sensors can acquire rich information under low-light conditions and greatly enhance night observation capabilities in the VNIR spectral range. Our country is catching up in this field. Domestic low-light satellites operate on the early-morning orbit, leading to great radiometric dynamics in the FOV and illumination variations become the dominant signal in images, so that image content details are severely suppressed. Accurate illumination correction is required before the data is used in visualization and quantitative applications. The current methods are either unpublished (GMA method for DMSP) or have unacceptable correction residuals near the terminator (NCC method for NPP), and therefore, better illumination correction method is needed for domestic low-light imagery systems. This project intends to accurately obtain ground radiation variations under low-light conditions by field measurements and to improve the low-light illumination correction methods based on the ground truth knowledge.

本项目拟提出一种改进的微光遥感图像光度校正算法，用以改善我国微光卫星的产品质量。近年来，以美国DMSP卫星的OLS载荷和S-NPP卫星的VIIRS载荷为代表的星载微光成像系统受到广泛地关注，它们能够在晨昏及夜间条件下获取可见光-近红外谱段的大量信息，极大加强了夜间观测能力，我国在此领域也在加紧追赶。我国微光成像系统运行在晨昏轨道，观测区域内辐射动态范围极大，光照强度变化成为图像信号主体，图像内容细节被压制，需要精确的光度校正处理，这是数据可视化及定量化应用的前提。国际现有的微光图像光度校正算法，存在技术不公开（DMSP）或者校正残余较大的问题（NPP的NCC产品），需建立更精确的微光图像校正方法和模型参数，以解决我国微光载荷应用之需。本项目拟采用野外测量与图像分析相结合的手段，准确获取微光条件下的地面辐射变化规律，并以此为基础改进微光图像光度校正算法。

本项目提出了一种改进的微光遥感图像光度校正算法，用以改善我国微光卫星的产品质量。我国微光成像系统运行在晨昏轨道，观测区域内辐射动态范围极大，图像内容细节被压制，精确的光度校正处理是数据可视化及定量化应用的前提。国际现有的微光图像光度校正产品校正残余较大，本项目通过采用专门设计的微光照度计，在敦煌定标场进行了为期8个月的连续野外测量，获取了大量的无人工光源干扰的自然辐照变化数据集，数据辐射的动态范围超过9个数量级，覆盖了从夏季到冬季及各种太阳、月亮角度和月相条件。通过对测量数据进行精细的辐射校正处理，消除了仪器各档位之间辐射不一致、像元饱和、天气等各种干扰因素，获取了微光条件下的地面辐射变化规律数据，并以此为基础改进微光图像光度校正算法的效果。与美国NPP卫星数据的同类产品相比，处理效果获得了明显的提升。