The Multimedia Signal Processing and Understanding Lab (MM-Lab) deals with most relevant aspects of multimedia data processing, focusing onto both theoretical and application-driven research issues. Since media can be considered as the technological extensions of human capabilities, particular attention is paid to smart multimedia data management, analysis, transmission and protection, and to those applications where ambient intelligence can provide advanced services beyond human limitations.
The Multimedia Signal Processing and Understanding Group is active in the research areas connected with the whole multimedia content lifecycle, from representation and coding, to processing, storage&retrieval, protection, analysis, and understanding. Major focus is on visual data, and in particular still picture and video. In this field, different fundamental aspects are considered, including: computer vision (visual tracking, action recognition, social interaction analysis, crowd analysis, smart camera repositioning), semantic media retrieval (content and context based indexing, event-based image and video understanding, gamification of media retrieval, social media), multimedia forensics (tampering detection, source identification, CG versus natural discrimination, anti forensics, digital watermarking).
Research at Multimedia Signal Processing and Understanding Lab always care about real-world applications. The attention to technology transfer is witnessed by several R&D projects developed by the group in the past years, and by the numerous links with local, national and international industries and research centers. At present, three major application areas are targeted. The first scenario concerns ambient awareness technologies for assisted living and e-inclusion. Here, automated monitoring systems are created that allow building applications in the field of videosurveillance, remote presence and domotics. The objective is to use multimedia technologies to acquire a complete knowledge about a monitored environment, and to react to events through automatic functions and advanced user interaction. In this framework, innovative sensing technologies such as wireless video sensors networks are also considered, with particular interest in lightweight coding and processing algorithms to limit computation and battery consumption. Furthermore, distributed processing paradigms are investigated to allow rapid deployment and easy configuration of video-based monitoring tools. Human-computer interaction is another challenging application field, which aims at developing advanced visual interfaces providing natural and intuitive interaction, and removing any technological barrier, which would limit the accessibility to the system and, especially, user acceptance. Specific interfaces have been designed that allow interacting with a system with the simple movement of hands or eyes. Experimental trials are being conducted in collaboration with medical doctors in the framework of rehabilitation and in design of innovative remote controls for elderly and disabled. The third application domain concerns the development of innovative multimedia content-and-context-based search engines. Here, the interaction with the user together with the exploitation of a-priori knowledge are exploited in order to bridge the gap between the current multimedia understanding algorithms and the richness and subjectivity of semantics in human interpretations of multimedia data. Data management includes also activities related to digital rights management and intellectual property preservation, digital content access control and manipulation detection. On the one hand, advanced cryptographic and watermarking techniques allow satisfying the required data security level, while, on the other hand, media content source authentication and image tampering detection can be provided by data hiding and digital forensics.
The Multimedia Signal Processing and Understanding Lab is currently exploring highly interdisciplinary initiatives, such as the Application Labs on E-inclusion and E-environment. The goal is to create collaborations among traditionally different research areas and foster new research communities on the boundary of the existing disciplines.
|Giulia Boato||Nicola Conci||Francesco De Natale (Coord)|