Using Pattern Noise of Imaging Sensors for Revealing Digital Forgeries
The method enables to decisively answer the following questions: Given a digital image and either the imaging device that took it or sufficiently many images from that device, was some specific area of the image tampered? Is there a tampered area in the image and where? As digital images and video continue to replace their analog counterparts, reliable and inexpensive content integrity verification of digital images increases on importance. It would especially prove useful in the court. For example, the integrity verification could be used for confirming the content of images presented as evidence, or, in a child pornography case, one could prove that certain imagery, or at least its critical part, has been obtained using a specific camera and is not a computer-generated image. The process of image integrity verification has been approached from several different directions, but to the best of our knowledge none so far led to a generally usable reliable method. The technology in this disclosure uses as an identification pattern a certain component of the pattern noise of imaging sensors (e.g., CCD or CMOS) caused by pixel non-uniformity. This pattern is approximately extracted using a denoising filter. The presence of the pattern in a given image area is established using a mathematical operation called correlation. This approach is computationally simple and relatively reliable. It is also possible to verify integrity of processed images (e.g., after JPEG compression or other common processing operations).
- High reliability and accuracy.
- Simplicity and computational efficiency.
- Applicability to all sensor types and image acquisition devices.
U.S. 8,160,293; 8,855,358
Binghamton University RB222