부산대학교 도서관

Hyperspectral images (HSIs) are the images that have a high spectral resolution with hundreds of bands. HSIs are widely used in the precise identification and classification of the materials and surfaces. However, they have a low spatial resolution caused by the limited capacity of the sensors responsible for capturing the images. In view of this, the preprocessing techniques are used to increase the spatial resolution of the HSIs and enable greater precision in the subsequent processing stages. This letter presents the development of a hardware accelerator specially designed to process a pansharpening algorithm that fuses the hyperspectral and panchromatic images to produce a high spatial resolution HSI. The experimental results demonstrate that the processor is energy-efficient and can be used for onboard processing in a small spacecraft, such as the satellites.