FPGA for Computing the Pixel Purity Index Algorithm on Hyperspectral Images

The pixel purity index algorithm is employed in remote sensing for analyzing hyperspectral images. A single pixel usually covers several different materials, and its observed spectrum can be expressed as a linear combination of a few pure spectral signatures. This algorithm tries to identify these pure spectra. In this paper, we present a Field Programmable Gate Array implementation of the algorithm, which has been implemented on a Virtex-II PRO XC2VP30 and on a Virtex-4 XC4VFX60 FPGAs and evaluated using the well-known “Cuprite” image (a standard benchmark in hyperspectral imaging applications). Our experimental results demonstrate that a hardware version of the PPI algorithm can significantly outperform an equivalent software version of the algorithm and retain excellent pure spectral extraction accuracy. In addition, the proposed architecture is easily scalable depending of the available resources and is more than three times faster than a recently developed FPGA implementation of the same algorithm due to the architectural improvements.