Book Description
New multimedia database search tools require rapidity, compression, image quality and copyright control. The two latter are often covered by embedding a hidden watermark signal in the multimedia content. More effective and recognized multimedia search methods are using hash functions. Hash functions compute very short signatures characteristically very different from one media to another. Typical requirement of hash functions are: easily computable, provide short output bit length, and collision resistant. In this thesis, an image hashing method for multimedia contents is presented as an innovative solution for content identification and indexing. This new one-way function for images extract some image features by different computations along radial projections. The magnitude of change is related to the amount of the change of the media (image, video). In watermarking, hashing enables the creation of payloads that depend on the media content, and which are thus resistant to the "copy attack" and collusion. This thesis is available in the SIMILAR collection under the label « Presses universitaires de Louvain » (ISBN 2-930344-58-X) and may be purchased online at www.i6doc.com.