Spatial and Spectral Resolution Limits of Hyperspectral Imagers Using Computed Tomography: A Comparison

Recently, a new type of hyperspectral imaging sensor has been proposed which simultaneously records both spectral data and multiple spatial dimensions. Dispersive imaging spectrometers typically measure multiple wavelengths and a single spatial dimension. Unlike dispersive imaging spectrometers, chromo-tomographic hyperspectral imaging sensors (CTHIS) record 2 spatial dimensions as well as a spectral dimension using computed tomography (CT) techniques with only a few spatially-spectrally diverse images. Different CTHIS designs have been proposed using different chromatic dispersion elements (such as prisms or diffraction gratings [2][4] [3]), but to date no comparison has been performed. In this paper, we examine the two design proposals for CTHIS and use the Cramer-Rao theorem to determine the lower bound of the variance of an unbiased object estimator of two point sources at different wavelengths. This lower bound will be useful to determine which sensor has the best spatial and spectral resolution in the presence of Poisson noise.