Reconstruction Of Fraction Surfaces From Data With Corrupted Pixels

Spectral unmixing is a key tool for a reliable quantitative analysis of remotely sensed data. The process is used to extract subpixel information by estimating the fractional abundances that correspond to pure signatures, known as endmembers (EMs). In standard techniques, the unmixing problem is solved for each pixel individually, relying only on spectral information. Recent studies show that incorporating the image’s spatial information enhances the accuracy of the unmixing results. In this chapter, we present a new methodology for the reconstruction of the fraction abundances from spectral images with a high percentage of corrupted pixels. This is achieved based on a modification of the spectral unmixing method called Gaussian-based spatially adaptive unmixing (GBSAU). Besides, we present a summarized review of the existing spatially adaptive methods.