The accuracy of spectral image reconstruction from threeband images can be improved by utilizing multipoint spectral measurements as auxiliary information. There can be different approaches for exploiting the multipoint spectral data. In Wiener and GMD (Gaussian mixture distribution)-based estimation methods, spectral data are employed to estimate the spectral statistics of a scene. The information of spatial coordinates of each spectral data can be also used for instance, with using the spatio-spectral MAP (Maximum a posteriori) estimation. However, spatio-spectral MAP requires a relatively large amount of computation.In this paper a novel method is presented as an alternative approach to make use of the spatial information of multipoint spectral data, called piecewise Wiener estimation. Then, we make a comparison of four methods: spectral Wiener, GMDbased, spatio-spectral MAP, and piecewise Wiener estimation algorithms. As a result, it has been found that methods that utilize the location information of each spectral measurement give higher accuracy than others.
Yuri Murakami, Kunihiko Ietomi, Ayumi Tadano, Masahiro Yamaguchi, Nagaaki Ohyama, "Comparison of spectral image reconstruction methods using multipoint spectral measurements" in Proc. IS&T CGIV 2008/MCS'08 4th European Conf. on Colour in Graphics, Imaging, and Vision 10th Int'l Symp. on Multispectral Colour Science, 2008, pp 591 - 596, https://doi.org/10.2352/CGIV.2008.4.1.art00127