Plenary Lecture
Digital Image Authentication Methods
Professor Mariko Nakano-Miyatake
National Polytechnic Institute of Mexico
Mexico
E-mail: mnakano@ipn.mx
Abstract: Nowadays a huge amount of digital images, with or without commercial value, are easily shared among the general public via Internet or stored using any of the several available digital formats. Such images which include private pictures or confidential images have in general high quality and may be used as important evidences of some accidents, illegal or criminal acts, etc. In general, such images can be easily manipulated using very efficient tools producing images that look natural enough, without any perceptible visual artifact, such that they can be easily assumed as authentic. Considering that these modifications may destroy important legal evidence or cause important economical or moral damages, efficient authentication techniques for digital images are required. Different image-authentication techniques have been proposed, which can be sorted into two types: watermarking-based and image hashing-based schemes, both of them with advantages as well as some drawbacks. Watermarking-based techniques insert an imperceptible signal into the image to be authenticated to create a watermarked image. The embedded signal can be a random signal or a signal related to the image to be authenticated. During the authentication process, the watermark is extracted from the watermarked image to be utilized for authentication, or even to restore the tampered image. The watermarking-based scheme performs fairly well and even allowing in some cases to restore the tampered image, although in some cases it may introduce some perceptible distortion in the image to be authenticated. The developing of efficient watermarking algorithms have been a topic of active research and several high-performance methods for embedding information into digital images have appeared in the literature. Otherwise the image hashing-based techniques, also called as multimedia fingerprinting, take out a set of robust features from the image to be authenticated to create a compact code, which is stored or transmitted separately, to be used for authentication. During the authentication process, this code is extracted from the image under analysis employing the same method used to estimate the stored or transmitted authentication code, which is then compared with the code extracted from the suspicious image and if the difference between both codes is smaller than a given threshold, the suspicious image is considered as authentic; otherwise, it is determined as a tampered image. So in the image hashing-based technique, the authentication code (hash code) must satisfy the following conditions: first the authentication code and the code extracted from the suspicious image must be quite similar if both images are perceptually similar, while the codes extracted from perceptually different images must be quite different. The second condition is that the authentication code must be as compact as possible, because it is necessary to reduce the storage space or transmission bandwidth caused by the authentication code. It is necessary to point out that the hashing technique or multimedia fingerprinting is different from the cryptographic hashing as in the last one, any change in the image to be authenticated, even if it is perceptually similar to the original one, produces a quite different hash value. In this talk is presented a review of some successful image watermarking-based and perceptual-hashing based image authentication schemes with tamper detection capability.
Brief biography of the speaker: Mariko Nakano-Miyatake received the M.E. degree in Electrical Engineering from the University of Electro-Communications, Tokyo Japan in 1985, and her Ph. D in Electrical Engineering from The Universidad Autonoma Metropolitana (UAM), Mexico City, in 1998. From July 1992 to February 1997 she was a Department of Electrical Engineering of the UAM Mexico. In February 1997, she joined the Graduate Department of The Mechanical and Electrical Engineering School of The National Polytechnic Institute of Mexico, where she is now a Professor. Her research interests are in information security, image processing, pattern recognition and related field. Dr. Nakano is a member of the IEEE, RISP and the National Researchers System of Mexico.