Video Fingerprinting Algorithm

Increasing amounts and various formats of multimedia data require technologies to automatically identify its content. Cryptographic hash functions can not be used for multimedia content identification or verification as they are sensitive to common video processing operations e.g. compression, colour adjustment etc. In this case, fingerprinting, also called perceptual hash functions, that consider perceptual similarity are applicable. Fingerprinting converts large data to a short digest, which is used to identify original data. This book describes some of the different existing approaches for video data. One reference implementation based on Philips algorithm is investigated. The fingerprints are extracted from the spatio- temporal differencing of mean luminance. It is shown how this method can be improved by using a simple similarity measure. The performance of the algorithms are analysed. The algorithm using similarity shows increased reliability of video identification both in robustness and discriminating capabilities.