Like actual fingerprints, the digital "noise" in original images is stochastic in nature – that is, it contains random variables – which are inevitably created during the manufacturing process of the camera and its sensors. This virtually ensures that the noise imposed on the digital images from any particular camera will be consistent from one image to the next, even while it is distinctly different.
In preliminary tests, Fridrich's lab analyzed 2,700 pictures taken by nine digital cameras and with 100 percent accuracy linked individual images with the camera that took them.
Binghamton University researchers have come up with a way link digital images to the camera that took the pictures and to detect forged images. The technology can be applied to prosecuting child pornographers. The technology is based on the fact that the original digital image is overlaid by a weak noise-like pattern of pixel-to-pixel non-uniformity, according to the press release. The human eye can't detect the patterns, the data can be extracted from cameras by analyzing several images taken from a camera to establish the digital fingerprint. Here's more from the release: