The Interference Imaging Spectrometer (IIM) onboard the Chang'E-1 satellite was used to retrieve the content of elements of the moon. The hue of IIM spectral images is inhomogeneous from left to right, which affects the retrieval of elements. Therefore, flat-field correction is necessary to eliminate the effects of the inhomogeneous spectral response of each pixel of the detector on the image. The IIM obtained a total of 576 orbits data during the one-year operation of the Chang'E-1 satellite (October 2007 to October 2008). Scholars have done flat-field correction before, but the correction results could not be applied to all regions of the moon. This paper combines the 130 orbits data obtained in November and December 2008, and proposes the reflectance normalization method based on spectral features. By selecting a large number of homogeneous, flat, and non-overexposed spectral data in beta-angle symmetrical orbits, the average row is obtained by averaging the selected data, and using the correction factors which are obtained by the average row to finish flat-field correction. Comparing the spectral images and reflectance curves before and after the correction, the inhomogeneity of hue and streaks in different terrains of the moon are effectively eliminated, which will be beneficial to the subsequent elemental retrieval and practical application.