• descrittori
• feature
• immagini iperspettrali
• registrazione
• SIFT
• SURF

In this work two of the main feature extraction techniques, the Scale Invariant Feature Transform (SIFT) and the Speeded Up Robust Feature (SURF)
algorithms, were analyzed and compared.
Such procedures were originally developed to work on gray level images. In this work we extended the SIFT and SURF procedures to hyperspectral images (HSI) so as to exploit the potentially
different information contents arising from different spectral bands.
In particular we applied the above mentioned techniques to multitemporal HSI registration task. The experimental analysis conducted on different real
hyperspectral dataset with very high spatial resolution highlighted the
effectiveness of the approach, resulting in a consistent improvement of the
registration algorithms performance.