Figure 3.

Schematic of the process by which a hyperspectral image is transformed into separate SWS and MLWS effective images. (A) The external scene is represented by a hyperspectral image “cube,” which has a single image at each of the 31 spectral bands. (B) An isolated image at a single spectral band/wavelength. (C) Using the blur circle calculated in Figure 2 for this wavelength and this retinal location, we used convolution to calculate the image on the retinal surface at this wavelength. (D) Normalized absorptance curves for the SWS cones and the MLWS cones (average of the MWS and LWS cones; see text for details). At this wavelength we determined the weighting factor for both cone types. Here, the SWS cone weighting factor was high, near 1.0, and the MLWS weighting factor was low, about 0.1. (E) Using the weighting factors, we add the image to the total summed SWS and MLWS images. The final result was two images: one representing the pattern of SWS cone activation across the retinal surface and one representing the pattern of MLWS cone activation across the retinal surface.