How Digital Photography Works

(^110) PART 2 HOW DIGITAL CAMERAS CAPTURE IMAGES
How Algorithms Smooth Out a
Picture’s Rough Spots
As soon as the newly digitized color
values come out of the analog-to-digital
converter, they are rushed over to
another chip, the array processor.
Its job is to guess the values of the miss-
ing colors based on the third of the val-
ues it does have and algorithms—a
cross between a formula and a plan of
attack. One such algorithm is bilinear
interpolation. To determine the miss-
ing green and blue values of the pixel
in the center, now occupied by only
red, bilinear interpolation averages the
values of the four nearest green and
blue pixels.
2
If we could look at a photograph before it was
processed and otherwise manipulated by the
camera, it would look something like this. It’s a
red, blue, and green mosaic, which helps
explain demosaicing; it means becoming some-
thing other than a mosaic. (We faked this photo,
by the way. Just so you know.)
1
In the creation state, while each photodiode was soaking in the rays of either red, blue, or
green light colored by the filter hanging above it, the diode was ignoring the other two colors.
That means two-thirds of the information needed to make up our photograph is missing! Of
course, there must be some scientific, high-tech methodology the camera can use to supply the
missing colors for all those millions of pixels. There is. It’s called guessing. But because scientific
types who design cameras don’t like to be caught making guesses, they have another name for
it:Interpolation. And on days they feel particularly self-conscious, they call it demosaicing.