The spatial-spectral compromise inherent in hyperspectral imaging (HSI) refers to one dimension of an HSI detector being devoted to spectral information instead of a second spatial dimension. By dividing light into contiguous bands of energy, HSI systems decrease recorded photon count per pixel from a scene often by a factor of 100 compared to typical RGB or panchromatic imaging systems, and by one to two orders of magnitude for their multispectral imaging (MSI) counterparts. This proves particularly problematic when imaging cultural heritage objects which may be sensitive to damage from intense illumination. A method to counter the signal loss is to sacrifice spatial resolution, increasing a pixel's projected footprint on a target, thus increasing the number of photons striking the pixel per time interval. Panchromatic sharpening is then applied in this research to visually recover the decreased spatial resolution. The increase in ground sample distance (GSD) can be accomplished physically, by increasing the imaging distance. Though this method does not increase the signal itself, as the projected area of the detector does not change, it does allow for quicker scan times over a given area. This allows the user to increase integration time, resulting in a higher integrated signal recorded over similar total scan times than at higher resolutions or lower imaging heights. In this research, a 14th century manuscript was imaged with an HSI detector under museum lighting levels of 50 lux. Here we show that as the spatial resolution was digitally downsampled by factors of approximately two and four, from 333 pixels per inch (ppi) to 161 and 80 ppi, the signal-to-noise ratio (SNR) was effectively increased by a factor of 2.00 and 2.79. Even after being spatially downsampled, the sharpened images were up to 1.5 times spatially sharper than the reference HSI image. Relatively high spectral accuracy was also maintained, with spectral angle mapper (SAM) measurements of 0.0527-0.0963.
|