Hyperspectral Image Visualization Based on Maximum-Reflectance Wavelength Colorization

Hyperspectral imaging is an important part of remote sensing technologies, providing detailed spectral infor-mation about the observed scene. Visualization of the resulting hyperspectral data cube is a challenge due to the large number of available spectral bands. In this paper, we propose a visualization technique that is based on mapping the corresponding RGB triplet to the wavelength value with the maximum reflectance. The approach is based on the assumption that the wavelength with maximum reflectance may reveal useful information about the objects and materials in the scene. We show experimental results on the widely-known Pavia University hyperspectral image. We interpret the results, provide a comparison with the existing methods and perform a quantitative evaluation for proving the usefulness of the proposed approach.