Method Research on Destriping Hyperion Image Noise

Hyperion is the first space-borne hyperspectral sensor in the world,which has abundant spectral information and spatial information.Hyperion has 242 bands.There are kinds of noise in Hyperion image due to different response of sensor and effection of absorption atmospheric such as water vapor,which will affect the interpretation and information extraction from Hyperion data.The characteristics of image noise were analyzed and 169 effective bands were selected.A right-left column mean method combined with median filter method was proposed to destripe the noise of Hyperion image.At last,the destriped image using new method proposed in this paper was compared with right-left sample mean method,globally destriping method,polynomial fit method.The results show that the new method can not only remove the striping noise effectively,but also remain more information from original image.