Extreme Pixels

It often happens that a circuit will have a few photoreceptors that are ``stuck'' at extreme values, either full on or full off. This comes about from several causes. An integrated circuit for the capture of images is likely to have many thousands or even millions of photoreceptors embedded in it, and will be exposed to a wide variety of conditions. Even if the probability of an individual photoreceptor being damaged is very small, the effect of having so many photoreceptors on a die will mean that there is a good chance of damage occurring. It should be remembered that there is a lens in front of the pixels, and therefore intense radiation, after being magnified, can easily cause permanent damage to a photoreceptor. There is also the likelihood of dust or scratches on the lens, that may be interpreted as a black or white spot.

A human retina is also likely to suffer such effects, as biological systems tend to be far more liable to damage than electronic ones. The retina manages to adapt itself so that, as far as possible, such pixels have no effect. We can use our common sense to spot isolated pixels in an image, but incorporating this common sense into an automatic system can be very difficult. The simplest solution is smoothing, as described above. An adaptive system could also be incorporated into the circuit, which detects these problems and ignores them. However, this adds greater complexity to the circuit, and it would be preferable to use a more elegant solution to eliminate these pixels.

Matthew Exon 2004-05-23