A region of interest (ROI) was selected from the ISO 12233^[g] test chart^[d] as shown below. It was modified to facilitate noise analysis by adding a gray patch. This modified test chart was used to calculate MTF/MTF50, aliasing and noise on the four types of images viz. Bayer, RGB, RGBW and Opponent.

Slant edge was selected as ROI from the ISO 12233 test chart as shown below^[d]:

This was modified to create the test chart used for the study:

The ROI for MTF calculation and noise analysis are marked in the above figure. The numbers within the black, white and gray regions indicate the R/G/B values from 0 to 255

Once the RGB test chart was created, it was converted to a multi-spectral scene for ISET by considering the fact that all levels of gray have same strength across wavelengths. For example, 31 wavelengths were assigned a value of 128/255 for the gray region of the test chart. This multispectral scene was then used for the study.

MTF50 was calculated using ieISO12233 function in ISET^[f] which is based on the 'sfrmat' program written by Peter Burns^[e]. Given a slant edge, the function calculates MTF and gives the values for MTF50 and aliasing. Given below is an output for Bayer image.

As can be seen in the above figure, for the Bayer image, the luminance channel MTF is always lesser than the green channel MTF, something not seen in RGB or Opponent images.

Noise is calculated on the noise ROI in the modified test chart. The R, G and B of the final image was weighted and added. Then the image was cropped to the gray ROI and standard deviation and mean of this region was calculated to derive the noise figure.

Only exposure time is varied in the study