EO-1 Hyperion高光谱数据的质量评价

Image Quality Evaluation of EO-1 Hyperion Sensor

本文以影像的客观评价方法对扬州地区一景Hyperion影像的L1R数据进行质量评价。主要采用辐射精度、信息量、清晰度、信噪比等指标对图像质量进行分析,经过评价认为,影像获取时受到成像环境和天气的影响,可用波段数量可能产生一定范围的浮动。本文的影像存在44个未定标波段,25个受水汽影响波段,而SWIR130以后的波段中存在大量噪声。影像信息主要集中于VNIR和SWIR波长较短范围的约94个波段内,该范围内影像质量较高,能发挥高光谱分辨率优势,并为正确还原光谱信息提供可能。影像数据中存在的条带现象和辐射畸变经过一定处理去除后,在农业调查、监测、管理,森林覆盖、灾害预警,地质调查,找油以及海洋水色研究等领域将有良好的应用前景。

This paper conducts the quality evaluation of a Hyperion LIR image of Yangzhou area by the objective image evaluation methods. The main indexes chosen for observation are radiation precision, the amount of information, sharpness, signal to noise ratio （SNR） and so on. It turns out that there are 44 zero bands and 25 water vapor affected bands among the 242 bands of the Hyperion data. The indicators of the remaining 173 bands are relatively moderate, except for those located near the water vapor affected bands, whose SNR are relatively lower. For this reason, it is considered that the environment and atmosphere conditions while acquiring images would affect the exact number of the available bands. As the conclusion, the image quality of the VNIR bands is obviously higher than that of the SWIR bands, and the image quality descends as the wavelength increases generally. The VNIR and SWIR bands before Band 120 possess the optimal image quality while bands after the 130th are of less sense for applications. Real spectral information could be extracted from these bands. After removing the striping noise and the smile effect, Hyperion image data could be used in the fields of agriculture, forestry, geological investigation, petroleum finding, ocean water color research and so on.