Title :
Unmixing-based denoising for destriping and inpainting of hyperspectral images
Author :
Cerra, Daniele ; Muller, Rupert ; Reinartz, Peter
Author_Institution :
German Aerosp. Center (DLR), Oberpfaffenhofen, Germany
Abstract :
Unmixing-based Denoising exploits spectral unmixing results to selectively recover bands affected by a low Signal-to-Noise Ratio in hypespectral images. This paper proposes to apply this algorithm, which operates pixelwise, for the inpainting of corrupted pixels and the removal of drop-out artifacts in hy-perspectral scenes. The reported experiments are characterized by a low reconstruction error for the reconstructed spectra and a high visual quality of the processed images, and outperform state of the art methods in terms of reconstruction error.
Keywords :
geophysical image processing; hyperspectral imaging; image denoising; image reconstruction; remote sensing; drop out artifacts; hyperspectral image destriping; hyperspectral image inpainting; hypespectral images; reconstruction error; unmixing based denoising; Hyperspectral imaging; Image reconstruction; Materials; Noise reduction; Signal to noise ratio; Three-dimensional displays; Spectral unmixing; denoising; destriping; hyperspectral data; image restoration; inpainting;
Conference_Titel :
Geoscience and Remote Sensing Symposium (IGARSS), 2014 IEEE International
Conference_Location :
Quebec City, QC
DOI :
10.1109/IGARSS.2014.6947522