Contents

s_PupilSizeBlur

Calculate the effect of pupil size on blur in the human retina model developed by Hopkins and implemented in humanCore.

The human OTF that we calculate is based on two elements. The first is the Hopkins model of the human eye. This is an idealized eye that accounts for chromatic aberration, but it does not have all of the features needed to capture the imperfections of the constructed human eye.

Hence, we added a second factor that degrades image quality. This comes from empirical measurements made in Dave Williams' lab. This reduces the contrast passed uniformly across all wavelengths uniformly.

But the degradation that we use is not pupil size dependent. This is a significant limitation in the accuracy of the model. In fact, for an idealized eye as the pupil size grows, the linespread will sharpen because there is less diffraction to contend with. In the real eye however, the data go in the other direction. As the pupil size grows the eye's aberrations play a larger role and there is degradation of the linespread.

Hence, the model we use here and in humanCore isn't handling pupil radius correctly (demonstrated in the calculations). The data are not bad around 3mm diameter, but they are too good at 6mm diameter.

One way to fix this problem is to have the degradation factor we use from the Williams' data depend on the pupil diameter. We should do this by examining published linespreads and adjusting the variable in humanCore called 'williamsFactor' to be pupil size dependent. (Wandell, May, 2010).

See also: humanOTF, humanLSF, humanCore

Copyright ImagEval Consultants, LLC, 2010.

if ~exist('isetbioRootPath','file')
    warning('Requires ISETbio.  Skipping.')
    return;
end

Warning: Requires ISETbio.  Skipping. 

We have a problem

pupilDiameterMM = 6;
oi = oiCreate('wvf human',pupilDiameterMM);  % The Thibos human model with 6 mm pupil diameter
oiPlot(oi,'psf',[],550);
set(gca,'xlim',[-10 10],'ylim',[-10 10]);

pupilDiameterMM = 3;
oi = oiCreate('wvf human',pupilDiameterMM);  % The Thibos human model with 6 mm pupil diameter
oiPlot(oi,'psf',[],550);
set(gca,'xlim',[-10 10],'ylim',[-10 10]);

Below here does not seem right to BW

Create linespread functions for a very large and a slightly small pupil

p1 = 0.003;    % 6 mm diameter
p2 = 0.001;    % 2 mm diameter
unit = 'um';
wStep = 5; wave = 400:wStep:700;
dioptricPower   = 59.9404;

% Set up LSF arguments
lsfBigPupil     = humanLSF(p1,dioptricPower,unit,wave);
[lsfSmallPupil,xDim,wave] = humanLSF(p2,dioptricPower,unit,wave);

% You can re-run this for various wavelengths by re-selecting w.  The
% wavelength range is from 400:700 in 1 nm increments
for thisWave = [450,550]
    fig = vcNewGraphWin;
    idx = (thisWave - wave(1))/wStep + 1;
    plot(xDim,lsfSmallPupil(idx,:),'r--',xDim,lsfBigPupil(idx,:),'b-');

    xlabel('Spatial position (um)');
    ylabel('Relative intensity');

    legend(sprintf('Diam = %.1f mm',2*p1*10^3),sprintf('Diam = %.1f mm',2*p2*10^3));
    title(sprintf('Line spread %.0f nm',wave(idx)));
    grid on

end

% Note that the LSF of the larger pupil size is sharper than the smaller
% size.  This is not right.  See comments at the top for how we should
% adjust this.

Published with MATLAB® R2024a