Multiple exposures

Multiple capture (e.g., bracketing) are common. Here is an example of combining multiple exposures in ISETCam.

Wandell, 2019

See also sensorComputeMEV, t_sensorBurst

Set up the default scene and the oi
Run the sensor simulation
Now in MEV mode
One combination method
Notice that there is less noise in the dark regions
For comparison with multiple exposure
This is the same computation but using the sensorComputeMEV method
END

ieInit
wbState = ieSessionGet('waitbar');
ieSessionSet('waitbar',false);

Set up the default scene and the oi

% We will use an HDR scene before long, rather than this default.
% scene = sceneCreate;
scene = sceneFromFile('Feng_Office-hdrs.mat','multispectral');
% sceneWindow(scene); sceneSet(scene,'display mode','hdr');
oi = oiCreate; oi = oiCompute(oi,scene);

Reading multispectral data with mcCOEF.
Saved using svd method

Run the sensor simulation

% First in single auto exposure mode
sensor = sensorCreate;
sensor = sensorSet(sensor,'fov',sceneGet(scene,'fov'),oi);
sensor = sensorCompute(sensor,oi);
sensor = sensorSet(sensor,'name','Auto exposure');
sensorWindow(sensor);
sensorSet(sensor,'gamma',0.3);

ip = ipCreate;
ip = ipCompute(ip,sensor); ipWindow(ip);
rgb = hdrRender(ipGet(ip,'srgb'));
figH = ieNewGraphWin([],'wide');
subplot(1,2,1); imagescRGB(rgb); title('Auto exposure')

Now in MEV mode

You might find a maximum time this way. expTime = autoExposure(oi,sensor);

maxTime = 0.2;   % seconds

% Simple bracketing case.  Increasing order of equally scaled exposure
% times.
expTimes  = [maxTime/10, maxTime];
sensorMEV = sensorSet(sensor,'exp time',expTimes);

% Get the multiple exposures.  This will display in bracketing mode
sensorMEV = sensorCompute(sensorMEV,oi);
sensorMEV = sensorSet(sensorMEV,'name','Multiple');
sensorWindow(sensorMEV);

One combination method

% Get the voltages from the multiple exposures
volts      = sensorGet(sensorMEV,'volts');
vSwing     = sensorGet(sensorMEV,'pixel voltage swing');
nExposures = sensorGet(sensorMEV,'n exposures');
expTimes   = sensorGet(sensorMEV,'exp time');
maxTime    = max(expTimes(:));

% Start out with the volts from the longest exposure time
combinedV = volts(:,:,end);

% Find the saturated pixels and replace them with estimates from
% shorter durations.  These may be saturated, too.  So, we loop down until
% there are no saturated pixels
thisExp = 1;
maxV = vSwing*0.95;
for thisExposure = (nExposures - 1):-1:1
    % Find the saturated pixels for this level
    lst = (combinedV > maxV);
    fprintf('Exposure %d.  Replacing %d pixels\n',thisExposure,sum(lst(:)));
    if sum(lst(:)) == 0, break
    else
        % Scaled volts for the shorter duration.
        sFactor = (maxTime/expTimes(thisExposure));
        theseV  = volts(:,:,thisExposure)*sFactor;
        combinedV(lst) = theseV(lst);
        maxV = maxV*sFactor;
        thisExp = thisExp + 1;
    end
end

Exposure 1.  Replacing 4499 pixels

Notice that there is less noise in the dark regions

sensor = sensorSet(sensor,'pixel voltage swing',max(combinedV(:))/0.95);
sensor = sensorSet(sensor,'volts',combinedV);
sensor = sensorSet(sensor,'name','combined');
sensorWindow(sensor);

For comparison with multiple exposure

ip = ipCompute(ip,sensor); ipWindow(ip);
rgb = hdrRender(ipGet(ip,'srgb'));
figure(figH);
subplot(1,2,2); imagescRGB(rgb); title('Multiple exposure')

This is the same computation but using the sensorComputeMEV method

maxTime  = 0.2;   % seconds
expTimes = [maxTime/10, maxTime];
sensor   = sensorSet(sensor,'exp time',expTimes);

sensorMEV = sensorComputeMEV(sensor,oi);
sensorMEV = sensorSet(sensorMEV,'name','Multiple exposure 2');
sensorWindow(sensorMEV);

ieSessionSet('waitbar',wbState);