Spectrum shape elements model to correct color and hyperspectral images

This paper presents a novel model of spectrum shape elements for correction of non-isoplanatic deviation in illumination. The model is applicable for color and hyperspectral images. We propose a technique for identification of the correction transform on the set of spectrum shape elements with the Hausdorff metric. In addition, a necessary condition, which allows obtaining an adequate form of the color correction transformation, is presented and theoretically proved. The experimental testing was performed on real color images and AVIRIS hyperspectral remote sensing data. The results show high quality of the proposed color correction technique.