An interesting article from the Discovery Channel about using noise to match digital images to the camera that shot it in much the same way bullet markings can be used to match to the gun that fired it.

My Nikon D70 has a really hot pixel in the lower left of the sensor (upper right). And D70’s in general on long exposures have a large amount of dark current digital noise in a distinct area caused by off-chip processing unit heating up.

The only problem I see is when one applies statistical noise reduction processing on the image (a la Neat Image or NoiseNinja), it should make noise identification very difficult. Heck, even in-camera noise reduction should make this hard? I wonder the impact of long exposure NR in my camera.

I hope the paper covered this.

So many questions…