Detecting the Adjacency Effect in Hyperspectral Imagery With Spectral Unmixing Techniques

The adjacency effect is an interesting phenomenon characterized by the occurrence of path interferences between the reflectances coming from different ground-cover materials. The effect is caused by atmospheric scattering, hence a typical approach to its detection has been the modeling of radiation transfer and spectral correspondence at particular wavelengths. In this paper, we investigate the detection of adjacency effects as being a general unmixing problem. This means that we opt to use spectral unmixing to separate the true signature of a pixel from the background scatter reflected from its adjacent neighborhood. To account for different types of atmospheric scattering, we consider several unmixing methods. These include the established linear- and a recently studied generalized bilinear model, as well as a more data-driven unmixing that could implicitly address nonlinearities not covered by the first mentioned approaches. We evaluate these unmixing models by comparing their results with those obtained from a specialized treatment of the adjacency effect in turbid waters surrounded by vegetated land. This comparison is demonstrated on real data acquired under varying atmospheric conditions.