ne of the best things about digital photography is the instant feedback. Snap a picture and you’ve got a 1x2-i nch p review of the imag e you j ust took, and m inutes later you can view that image on a monitor or make a pr int. Comp are this turnarou nd to the hours or days required to get prints or slides from film. This rapid feedback is terrific for l earning how to take great pictures in unu sual conditions. F or instance, infrared and night p hotog- rap hy each require some p racti ce. The downside to night ph otography w ith digital cameras is dark current noise. I n ad dition to higher grain, long exposures tu rn certain CCD el ements on , gi ving your picture of the n ight sky a few extra stars! F ortun ately , with a littl e care, this problem can be greatly redu ced. Below you see the original noisy image on the left and 30 • PEI• JANUARY 2001 Eliminating Dark Current Noise in Long-Exposure Digital Images Left: The original image (16 second manual exposure, f/2.8, 100 ISO equivalent) shows dark current noise. Right: the final image after adjustment.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
natu ral way to fix this is to simp ly subtract these pixels.
Take the base image, hold d own the shift key and drag
the DCN correction image on to it to create a new layer,
and select the Difference mod e in the Layers p alette.
Now the noise in the dark sky an d trees is gone.
But there’s a dow nside to d ifferencing. Subtracting
the noise works w ell for astronomers looking at stars at
night. Unfortunately for photographers, subtraction
creates the opposite problem of lit-up pixels: dark sp ots
now show u p in bright p arts of the image (look at the
refinery lights). Dark current noise occur red even in
these bright areas of the image. Although we d on’t
wan t noise in the d ark areas of the picture, we need to
keep it in the bright p arts!
Adding Things UpWhat w e really want to d o is not subtract off the
noise, but to rep lace noisey pixels with the better ones.
This leads to ou r
final approach,
combining these
two m ethods. We
use blurring to
generate good d ata
for the replacement
pixels, and the
noise-only image to
control which pixels
to replace.
First we need to
get good replace-
ment p ixels. Take
the base image,
du plicate it into anoth er layer, and heavily blur th at
layer. (The exact blur rad ius depend s on image
resolution, but don’t be shy. I used Radius=5.0.) No
more noise here. Nam e this layer “DCN rep lacement.”
Now open the DCN correction image in another
window. When you op en it, do not allow any color
management; the raw d ata is needed in the n ext steps.
We’ll use it to construct a mask to choose which pixels to
replace. Convert it to grayscale (Image>Mod e>Grayscale).
Typically on ly a few elements are comp letely blown ,
but m any w ill be partly corrupted . We need to replace
all the pixels, so turn up the gain (Image>Adjust>
Levels; click “Auto”). Make sure the output levels are
scaled 0 to 255 and that m ost of the image remains
black. This image w ill serve as the m ask for the DCN
replacement layer, so select it all and copy it (this will
be the “DCN m ask”). Go back to the base image an d
select the “DCN replacement” layer in the Layers
dialog, add a m ask (Layer>Add Layer Mask>Reveal
All). Then paste the DCN layer into the mask (op t/ alt-
click on th e mask and paste it).
As a final step, you m ay wish to app ly a light
Gaussian b lur (Radius=0.5 pixel) to the image, wh ich
will remove some of the remaining grain.
You’re now set to take long-exposu re pictures at
night.
John Heidemann is a research assistant professor of computerscience at the University of Southern California. Beforejoining USC, he developed image processing software at theUniversity of Nebraska-Lincoln Remote Sensing Center.
32 • PEI • JANUARY 2001
Paste the dark image over the original image and selectDifference in the Layers palette to eliminate noise in the dark areas.
Turn up the gain in your DCN correction image to use it as a mask.
Use a very blurred image layer foryour replacement pixels.