Data Embedding And Digital Watermarking

LNT I Seminar Winter 1998 Data Embedding and Digital Watermarking Jonathan K. Su Telecommunications Laboratory I University of Erlangen-Nuremberg Outline ● ● ● ● ● ● Motivation Desired properties Spread-spectrum concepts Watermarking examples Attacks and robustness Conclusions J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 2 Digital Documents and Data Embedding ● ● ● Examples of digital documents: formatted text, audio, images, video, 3-D objects, VRML, etc. Data embedding: “Perceptual headroom” allows additional information to be inserted directly into a document. Digital documents can be stored, copied, and distributed easily, rapidly, and with no loss of fidelity. J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 3 Data Embedding Applications ● ● ● ● Provide supplementary information or additional functionality Covert communication Track usage, modifications, distribution Watermarking: Protect intellectual property rights (IPR, e.g., copyrights) J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 4 IPR Protection: Why Use Digital Watermarking? ● ● Existing mechanisms ❍ Encryption ❍ Copy protection Digital watermarking ❍ embed information about origin and recipient directly in document ❍ does not prevent copying, but allows tracking ❍ “last line of defense” J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 5 IPR Protection: Watermarks as Fingerprints ● Identification of legal recipient with help of individual watermarks: .... Originator watermark B1 watermark A watermark B2 watermark B3 .... Digital Library (e.g. WWW Server) J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. illegal copy 6 Desired Properties: Imperceptibility ● ● After embedding, document should remain perceptually equivalent to original. Perceptual models may be employed. Original After embedding J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 7 Desired Properties: Robustness ● Embedded information cannot be damaged or destroyed without making the document useless. JPEG compression Additive noise & clipping J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 8 Desired Properties: Security ● ● Information cannot be detected, read, and/or modified by unauthorized parties. Security resides in the secrecy of the key, not in the secrecy of the algorithm. document + embedded information correct key incorrect key Recovery J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. correct information garbage 9 Additional Desired Properties ● ● ● ● ● (No) reference to original document during retrieval ❍ interference from original Multiple embedding ❍ one copy with several information streams ❍ unique information for different copies Fast information embedding and/or retrieval Compressed-domain processing Bit-rate constraints J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. 10 A Communications Model Message (information) EMBEDDING Watermark generation (Perceptual model?) Key Received message Message recovery RECOVERY Watermark (signal) Watermark insertion Original document Watermark and remaining interference J. Su, Data Embedding and Digital Watermarking, LNT I Seminar, WS 98. Marked docu