利用多项式拟合方法去除Hyperion数据条带噪声

Destriping in hyperion data based on least-square polynomial fit

Hyperion是目前唯一在轨的集高空间分辨力与高光谱分辨力于一体的成像光谱仪,Hyperion数据含有丰富实用的光谱信息。由于各传感器光谱响应值不一致,造成许多谱段含有大量条带噪声,严重影响Hyperion数据的解译和信息提取。介绍了常用于Hyperion数据的去除条带噪声方法,提出了采用多项式拟合的算法,并对各方法去条带效果进行比较。实验结果证明,多项式拟合算法优于其他方法,去条带效果明显,且能很好地保持原始图像特性。

EO-1 Hyperion sensor which provides high spatial and spectral earth observation data will become more readily avaliable to the research and user communities.The column striping is attributed to variable respondence of the detector array.The striping noise can distractingly and obstructively affect the interpretation and application of Hyperion data.Some methods previously used in striping removal of Hyperion are introduced,and a new method using a least-square polynomial fit to interpolate the stripes is presented.According to the experimental results,the new method can achieve a better result than other methods mentioned in this paper in removing striping noise of Hyperion data,and it can preserve the spectral characteristic of original image.