Dr Ivan Damnjanovic

Contact Details

Research Group: Multimedia and Vision Research Group

Research Areas: Cross-media retrieval, A/V synchronization

Research interests: Cross-media retrieval, Digital Watermarking, A/V Coding, Perceptual Modelling, Information Theory, Error Correction Coding, Semantic Web.

Participation in projects: EASAIER , MESH , K-Space , BUSMAN

Cross-media retrieval in the EASAIER Project

One of the objectives of the EASAIER project is to enable cross-media retrieval in music archives. Metadata and extracted features are a valuable basis for cross-media retrieval. Metadata can be used to find in the database and rank all documents related to the song and the authors’ names. The ESAIER system will present with highest relevance the audio file containing the song, but ranking of other related documents remains challenging problem. The features extracted from documents will be used to enhance this task. A video that is related to the song can be presented to the user with high relevance if it contains a segment with a performance of the song. This can be found by computing similarity between the audio segments of the video and the audio recording of the song. In the similar way, computing visual similarity on keyframes can identify a video clip of an interview with the performer in which the same video segment is used.

Due to the combined use of metadata and features, the database must support several radically different types of internal search methods. First, relationships based on features require a multidimensional similarity index. There is a large literature on this but it remains to be seen which index is most suited for such a problem, and what modifications it would require. Furthermore, the metadata gives rise to complex relationships between documents. Ranked linkages of metadata connections may give rise to nonmetric relationships (an image may come from a film, which features a song, but the image is only tangentially related to the song). Thus the appropriate way to search the metadata remains a challenging task. Graph theory and small world networks may be highly applicable to this problem.

A/V synhronization

One of the main advantages of the enriched access system, developed in the EASAIER project, will be Time Scale Modification with synchronized video. This kind of tool is highly required by music researchers. It will enable a user to slow down the video and see the playing technique of particular artists, while still hear them playing in original tonality. To the our best notion, there is still a lack of available tools with this capability. Tools such as mplayer (www.mplayerhq.hu) can change playback speed, but at the same time changes the pitch – similar to the analog type of players. Perhaps the most advanced online product is Chronotron Pro (www.chronotron.com), which offers independent pitch scaling and time stretching, but with a tremendous reduction in both audio and video quality. A hybrid real-time TSM approach has been integrated with video player. Synchronization is maintained using PTSs (Presentation Time Stamps). The audio clock is chosen as master and video presentation follows changes in audio, by duplicating or skipping the frames depending on slowing down or speeding up the audio playback rate.

PhD Thesis: "Watermarking of Mpeg-2 Video Streams"

Scope: A rapid expansion in digital technology, we are witnessing in past two decades, introduced huge benefits both for the industry and for the consumers. However, it also raises a number of questions regarding digital media copyrights. Digital media encryption can provide secure delivery of the content, but once the content is decrypted on a consumer side, it cannot prevent unlimited copping of the material. Digital watermarking, as a technique for embedding secret watermark directly to the digital content, came as possible solution to complement encryption, but also to a range of other interesting applications such as authentication, broadcast monitoring or data embedding.

Digital media are mainly transferred and archived in a compressed form. Due to performance constraints in terms of computing efficiency and degradation, re-encoding and watermarking in the pixel domain is not feasible. Hence, it is desirable to embed and detect watermark in compressed domain. Several techniques have been reported in the literature aiming at watermarking in the compressed domain and they are mainly based on spread-spectrum paradigm, where a watermark message is transmitted as a narrow-band signal via a video signal, which acts as a wide-band channel.

One of the main requirements in digital watermarking is that the watermark must be embedded in such way that it does not introduce visual artefacts to the host signal. Since the watermark power is bounded by perceptual visibility, watermark bits need to be spread by a large chip-factor in order to have reliable detection, which will decrease a number of watermarking bits that can be put in the same embedding window. Detection reliability can be increased by pseudo-random spreading of the watermarking bits and also perceptual watermarking models can be incorporated to increase the embedding power. In our technique, we used combination of Watson’ JND model ( A.J. Ahumada, H.A. Peterson and A. B. Watson, “An Improved Detection Model for DCT Coefficients Quantization”, ”, Proc. of SPIE, Human Vision, Visual and Digital Display IV, ed. B. E. Rogowitz, vol. 1913-14, 1993, pp. 191-201.) and improved JND model using block classification (X.H. Zhang, W.S. Lin and P. Xue, “Imroved Estimation for Just-noticeable Visual Distortions”, Signal Processing, vol. 85, no. 4, 2005, pp. 795-808).

In compressed domain watermarking, another limiting factor is that the video bit-rate must remain the same. Embedding in a particular coefficient is not allowed if it increases the bit-rate giving decrease in the power of the watermarked signal. The number of altered coefficients can be increased with an optimized bit-rate preserving scheme on macro-block level.

The watermarking process can be seen as a communication of secret message via a noisy channel, where noise is originated by the host video signal by itself and the attacks. Although, above mentioned techniques can increase the watermarking power and Signal to Noise Ratio of the watermarking channel, the SNR is still low to communicate desirable number of bits with high decoding rate. Hence, it is necessary to introduce some form of error correction coding to boost the capacity and improve the detection rates.

We are proposing to use duo binary Turbo codes to protect watermarking channel. Turbo codes (TCs) have received great attention since their introduction in 1993. This is due to the extra-ordinary performance at low bit error rates, reasonable complexity, and versatility for encoding blocks with various rates and sizes.

Publications

Conferences

• F. Scharffe, M. Luger, Y. Raimond, I. Damnjanovic, and J. Reiss, "Semantics and Ontologies for Multimedia Objects Representation and Metadata Management in Sound Archives," in Proc. 4th International Conference on Automated Solutions for Cross Media Content and Multi-channel Distribution, AXMEDIS, 2008,
• I. Damnjanovic, D. Barry, and J. Reiss, "Enabling Access to Sound Archives through Integration, Enrichment and Retrieval," in Proc. Proc. of IEEE International Conference on Multimedia and Expo, 2008,
• I. Damnjanovic, C. Landone, P. Kudumakis, and J. Reiss, "INTELIGENT INFRASTRUCTURE FOR ACCESSING SOUND AND RELATED MULTIMEDIA OBJECTS," in Proc. 4th International Conference on Automated Solutions for Cross Media Content and Multi-channel Distribution, AXMEDIS, 2008,
• I. Damnjanovic, C. Landone, J. Reiss, and E. Izquierdo, "Enriched Access to Digital Audiovisual Content," in Proc. NEUREL, 2008,
• I. Damnjanovic and E. Izquierdo, "Perceptual Watermarking Using Just Noticeable Difference Model Based on Block Classification," in Proc. Proc. of MobiMedia 2006 - 2nd International Mobile Multimedia Conference, 2006,
• I. Damnjanovic and E. Izquierdo, "Capacity enhancement of Compressed Domain Watermarking Channel using Duo-binary Coding," in Proc. Springer Lecture Notes in Computer Science, International Workshop on Digital Watermarking IWDW2006, 2006, pp. 162–176.
• I. Damnjanovic, N. Ramzan, and E. Izquierdo, "MPEG-2 Watermarking Channel Protection Using Duo-binary Turbo Codes," in Proc. Proc. of IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 06, 2006, pp. 301–304.
• I. Damnjanovic, and E. Izquierdo, Turbo Coding Protection of Compressed Domain Watermarking Channel, Proc. of IEEE International Conference on Computer as a Tool, Belgrade, Serbia and Montenegro, 21-24 November 2005, vol. 1, pp. 171-174.
• I. Damnjanovic, and E. Izquierdo, “Capacity Enhancement of Compressed Video Watermarking Using Turbo Codes”, 6th European workshop on Image Analysis for Multimedia Interactive Services, Wiamis’05, Montreux, Switzerland, April 13-15, 2005
• Li-Qun Xu, Paulo Villegas, Mónica Díez, Ebroul Izquierdo, Stephan Herrmann, Vincent Bottreau, Ivan Damnjanovic, Damien Papworth, “A User-Centred System for End-to-End Secure Multimedia Content Delivery: From Content Annotation to Consumer Consumption”, CIVR 2004, pp. 656-664
• E. Izquierdo, I. Damnjanovic, P. Villegas, L.-Q. Xu, S. Herrmann, “Bringing user satisfaction to media access: the IST BUSMAN Project”, Proc. of Eighth International Conference on Information Visualisation, 2004. IV 2004, 14-16 July 2004 Page(s):444-449.