Unmixing low ratio endmembers through Gaussian synapse ANNs in hyperspectral images

In this paper we considered the application of Gaussian synapse based artificial neural networks to the detection and unmixing of endmembers in cases where some of them are mixed in a low ratio within hyperspectral images. These networks and the training algorithm developed are very efficient in the determination of the abundances of the different endmembers present in the image using a very small training set that can be obtained without any knowledge on the proportions of endmembers present. The validation and test of these networks is carried out through their application to a benchmark set of artificially generated hyperspectral images containing five endmembers with spatially diverse abundances. As a second test, we applied the strategy to a real image and checked their behavior in regions where there were transitions between zones that were labeled differently and compared them to a hypothetical evolution of the spectrum from the endmember corresponding to one of the regions to the endmember of the other. A very good correspondence was found.