A New Digital Repository for Remotely Sensed Hyperspectral Imagery on GPUs

Abstract

Hyperspectral imaging is a new technique in remote sensing in which an imaging spectrometer collects hundred of images (at different wavelength channels) for the same area on the surface of Earth. Over the last years, hyperspectral image data sets have been collected from a great amount of locations over the world using a variety of instruments for Earth observation. Only a small amount of them are available for public use and they are spread among different storage locations and exhibit significant heterogeneity regarding the storage format. Therefore, the development of a standardized hyperspectral data repository is a highly desired goal in the remote sensing community. In this paper, we describe the development of a shared digital repository for remotely sensed hyperspectral data, which allows uploading new hyperspectral data sets along with meta-data, ground-truth and analysis results (spectral information). Such repository is presented as a web service for providing the management of images through a web interface, and it is available online from http://www.hypercomp.es/repository. Most importantly, the developed system includes a spectral unmixing-based content based image retrieval (CBIR) functionality which allows searching for images from the database using spectrally pure components or endmembers in the scene. A full spectral unmixing chain is implemented for spectral information extraction, which allows filtering images using the similarity of the spectral signature and abundance of a given ground-truth. In order to accelerate the process of obtaining the spectral information for new entries in the system, we resort to an efficient implementations of spectral unmixing algorithms of graphics processing units (GPUs).