Content Based Identification (Fingerprinting)

Abstract

Ever since the broad availability of efficient source coding methods (data reduction) and digital distribution channels (including the Internet), consumers have seamless access to an enormous amount of multimedia data. This includes audio material and still and moving pictures within a wide range of quality, ranging from “pre-view” (e.g. Internet radio) to broadcast quality. As a result, efficient handling of this considerable amount of data has become a challenge of its own (e.g. “how can I find desired material efficiently?”). This has led to the definition of a number of so-called metadata standards. Examples for such specifications include the Dublin Core initiative, the SMPTE/EBU Dynamic Metadata Dictionary, the P/Meta project of the European Broadcasting Union (EBU) and, more recently, the MPEG-7 standard. The general idea behind these standards is to define data formats which provide a comprehensive description of the actual multimedia content in an interoperable way. Such meta-data (i.e. “data about data”) structures may include a wide range of descriptions of the origin and identity of the content, its structure, usage rules, and various perceptual or semantic aspects.