Doctoral meeting: 'Efficient registration of hyperspectral images on GPU'
The advances in sensor development in the last decades allow us to obtain multi- and hyperspectral images at a lower cost than before. Each pixel of these images contains a continuous spectrum that is formed by hundreds of narrow bands. As a result of this high spectral resolution, objects, plant species, and land–cover classes among others, can be distinguished with higher accuracy. Thanks to this large amount of available information, the use of multi- and hyperspectral remote sensing images has been extended to a multitude of applications such as vegetation science, land use classification, geology, quality control, and change detection among others.
A previous fundamental task in many of these applications is the registration of images of the same scene which have been taken at different times from different viewpoints and which, furthermore, present changes in objects, in illumination, etc. The goal of the registration is to estimate a geometrical transformation that maps one image to another. A similar problem arises when it is necessary to register the different bands of the same image.
The high dimensionality and the volume of hyperspectral information obtained by remote sensing sensors is something to bear in mind. This is specially true in applications with time constraints such as monitoring of chemical contamination. In this context, high-speed processing of common tasks such as registration is required.
In this thesis, we propose different methods for aligning multi- and hyperspectral remote sensing images by exploiting the available spectral information to improve the registration accuracy. They were developed for running entirely on GPU obtaining a considerable reduction in execution time compared to CPU implementations.