高光谱遥感图像的地形校正和评价方法

Terrain Correction and Evaluation Methods of Hyperspectral Remote Sensing Image

高光谱遥感具有光谱分辨率高,图谱合一的突出特点,在地质勘探和植被遥感等定量应用中,较全色和多光谱更有优势。然而目标区复杂的地形效应,是制约高光谱数据应用效果的诸多外界因素之一,利用地形校正模型消除地形影响,已成为目前高光谱遥感图像步入应用前的有效处理手段。目前的地形校正方法研究当中,遥感数据以多光谱居多,且校正效果的评价方法较之高光谱图像的特点也略显不足。通过朗伯体和非朗伯体假设的不同校正模型(C,SCS+C,Minnaert,Minnaert+SCS)对Hyperion遥感图像进行地形校正,并利用目视效果,分类标准差,目标反射率对比对校正结果分析评价,证明Minnaert+SCS模型最优,且评价方法合理,有效。

Hyperspectral remote sensing has prominent properties of high spectral resolution and the combining between image and spectrum. In the quantitative application of geological exploration and vegetation remote sensing, which has advantages than panchromatic and multi spectral. However, the complex terrain effect is one of the many external factors that restrict the application of high spectral data. Terrain correction models are used to remove terrain effect, and it has become an effective method for application of hyperspectral remote sensing images. At present, among the methods of terrain correction, the muhispectral images are used mostly, and the evaluation methods of the correction effect is slightly less due to the characteristic of hyperspectral image. In the paper, terrain correction for Hyperion image is used by different models（ C, SCS ＋ C, Minnaert, Minnaert ＋ SCS） with Lambertian and non - Lambertian assumptions, and the visual effect, classification standard deviation, target reflectivity contrast are used to analyze and evaluate the correction result. Above all, the Minnaert ＋ SCS model is best and the evaluation methods are reasonable and effective.