Research Activities in TUC/MUSIC
High Performance Multimedia Architectures
Multimedia data have demanding performance requirements due to their volume, bandwidth, and delay sensitive nature. This line of research aims to discover principles and algorithms for efficient multimedia management and delivery.
Multimedia Content Management Systems
This line of research aims to explore architectures and system implementations that support multimedia content creation, delivery with industrial standards and quality of service requirements, as well as user interaction support. The delivery can be done through internet lines, through satellites or through mobile infrastructures. A number of prototype systems have been implemented in cooperation with the European industry and used for demonstration and experimentation.
Semantics and Personalization in Multimedia Services
This line of research aims to provide advanced functionalities for the management of multimedia content. Mpeg-7/21 are the dominant international standards for multimedia content description such as images and video. We aim to provide methodologies for enriching them with semantics using ontologies while maintaining compatibility with Mpeg-7. We also provide a powerful framework for semantic personalization in Mpeg-7 and we explore the use of this semantic framework in all the stages of multimedia content processing, management and delivery.
Semantics in 3D Graphics and Spatial Applications
Traditionally 3D graphics software and spatial software (like GIS) are not well integrated with semantics. There are however several important applications where the semantics of the 3D objects or the spatial objects are important for retrieval, visualization, management, information capturing. This line of research explores systematic integration of ontologies in 3D graphics and Spatial data, and their use to provide advanced functionalities in applications that use them.
Semantic Natural Language Processing and Interfaces
Natural language processing and development of interfaces based on natural languages is important in many applications. When the content is restricted to a specific domain of knowledge better results can be obtained (less ambiguities arise) using domain ontolgies. In this line of research we developed methodologies for exploiting domain semantics in the natural language processing, and a software framework that facilitates the automatic generation of natural language interfaces to knowledge repositories. We are also pursuing research in the syntactic analysis of the Greek language.
Web Application Development Methodologies
This activity pursues research in the systematic development of large ubiquitous web applications from the requirements capturing phase to architectural patterns and frameworks for the development of large ubiquitous web applications. Intelligent ubiquitous collaborative interfaces using different media are also explored.
Semantic Web Services, Architectures and Systems
Web Services are emerging as a dominant platform for web application development. Semantic web services could utilize domain knowledge described in by specialized models and languages in a multilayer fashion (like MDA). SMEs will collaborate and compete in a peer to peer fashion, resembling a natural ecosystem. In this environment we develop models, architectures, languages and systems to support the SMEs in their work. Domain knowledge of the domain where the ecosystem exists are exploited to better support the functionalities required.
In web applications interoperability plays a particularly important role. We develop generic methods for semantic interoperability, for example interoperability between XML Schema Applications and OWL applications. We also develop semantic interoperability frameworks for particular classes of applications for example multimedia and cultural data applications, taking into account their supporting standards (Mpeg-7 and CIDOC/CRM).
Digital libraries explore different kinds of media and functionalities to support distributed access for specific applications like libraries, museums, learning, scientific, etc. Various research and development activities in providing functionalities for digital libraries are pursued.
E-learning applications are important in the knowledge society where people must be retrained several times in during their life time. We are pursuing a number of activities from system building exploiting the international standards of the field, to dynamic creation of learning material based on user preferences, to creating methodologies and systems for evaluation of e-learning applications, to creating in cooperation with experts learning material for specific application domains.
Culture and Tourism Applications
Culture and tourism applications are by nature multimedia applications that also have a strong need for spatial and mobile support. Different types of actors are involved (providers and users), and domain standards have to be followed. In this environment we have pursued research and development for ubiquitous collaborative applications in culture and tourism in cooperation with international and local actors.
Personalized, Interactive TV Applications
In an environment where thousands of TV channels are transmitting continuously programs, it is important that program selection and viewing is automated in order to work at all times (night, work hours) without the user supervision. We have developed applications that support personalized selection of programs from the programs broadcasted from various stations at all times. The user can view whenever he wants those programs and interact with the system for giving guides on what to do. The user profiles are automatically maintained.
This activity has generated in cooperation with large hospitals multimedia medical applications for supporting an integrated hospital environment, where the various departments of the hospital will have access for viewing information generated anywhere in the hospital or in other hospitals in a collaborative viewing environment.
Perceptually-based Real-time Selective Rendering Algorithms
In a Virtual Environment (VE), efficient techniques are often needed to economize on rendering computation without compromising the information transmitted. This research activity aims to devise a functional fidelity metric by exploiting research on memory ‘schemata’ related to the context of a scene (an office, a hospital, etc). According to the proposed measure, similar information would be transmitted between a synthetic and a real world scene, both depicting a specific schema. This would ultimately indicate which areas in a Virtual Environment could be rendered in lower quality without affecting information uptake.
Mathematical Modeling and Visualization of Archaeological Uncertainty
Archaeologists piece together available information derived from evidence into a speculative version of the past. This version becomes more certain as the evidence increases. For the past decade, 3D computer graphics archaeological visualizations have mostly been represented by photo-realistic reconstructions of ancient monuments and ruins. Because of the possibility of misleading the public, the archaeological community has stressed the need to acknowledge the availability of other possible hypotheses as well as the difference between what was found and how it is interpreted. This research attempts to mathematically model and visualize the archaeological expert's uncertainty in the visualization scheme. We define uncertainty as the extent of expert’s knowledge and confidence related to archaeological evidence which could ultimately be included in the visualization of an archaeological structure.
The Effect of Latency on Cognitive Tasks in Immersive Simulations
The proposed research is going to explore the perceptual as well as the cognitive impact of latency degradation on user perceptual thresholds and 3D spatial awareness in naturalistic spaces and meaningful situations. Since the combination of sensing, computation, rendering, and transmission delay (latency) is unavoidable in most Virtual Environments (VEs), simulations, tele-operation and augmented reality applications, interest is directed to how detectable differing levels of latency might be. Both the quantification of perceptual sensitivity to latency and description of the mechanism by which VE latency is perceived is essential to guide system design in the development of countermeasures such as predictive compensation. In this research, we examine the effect of latency of memory performance and memory awareness states.
Fidelity metrics for 3D immersive based on spatial perception and cognition
This line of research is exploring fidelity metrics for complex immersive simulations based on spatial perception (slant perception) and spatial cognition (memory awareness performance).
An innovative motoric measure of slant based on gait is proposed as, in effect, the angle between the foot and the walking surface during walking, exploring behavioural fidelity of simulations. This work investigates whether the proposed ecologically valid, action-based measure is affected by factors such as material of the walking surface and inclination of the walking ramps. Experimental studies were conducted in a real environment set-up and in its virtual counterpart.