This paper analyzed the importance of relative radiometric normalization(RRN) in remote sensing and aerophotogrammetry filed. The linear correction method of RRN was studied. A high automatic RRN method was proposed based on the scale-invariant feature transform(SIFT).It was found that the RRN can minimize radiometric difference among images caused by inconsistencies of acquisition conditions (such as weather, season, etc.) rather than in surface reflectance. Compared with the traditional ways, the new method is more robust and automatic.