Graphics: Visual and Interactive Computing

• VIC  | 
• Volume Graphics  | 
• Data Visualization  | 
• Interactive  | 
• Intelligent  | 
• Members

Intelligent Methods for Visual and Interactive Computing

Academics: M. Chen, P. W. Grant, M. W. Jones

Collaborators: Utah (USA), Swansea Theory Group (UK), industrial partners (UK)

Inspired by the recent developments in translation techniques for XML-based languages, the group formulated a novel approach to the automatic translation of graphical scene description languages using independent stylesheets. The principal concept of XSLT was further extended by replacing n(n-1) binary mapping stylesheets with n individual stylesheets, one for each scene description language. This approach is referred to as Independent Stylesheet Language Translation (ISLT), and it dynamically establishes the syntactic mapping and semantic approximation from one language to another using knowledge acquired from stylesheets and previous translation. A prototype was developed that successfully demonstrated the technical feasibility of ISLT.

Recently, the group developed a data-driven framework for facial aging simulation. In order to obtain a person-specific age-progression model based on an input image, genetic programming was employed to evolve a solution automatically by learning from example transformations in a facial image database. With evolutionary computing, this technique is able to infer from the input and the database the most appropriate models to be used for transforming the input face. The results obtained using this new approach represent a significant leap from those in the literature.

With the increasing complexity of large visualization jobs executing over distributed heterogeneous hardware, and displaying output on various devices (desktop PCs, PDAs), the management of such software is becoming difficult for both users and system integrators. We developed the approach of using software agents to manage visualization tasks, which inspect the running system, and then employ various strategies to improve the user response speed. These software agents also insulate the programmer from a lot of complex tasks as they provide strategies for parallelising rendering, migrating processes, obtaining user input and streaming output to multiple displays and display types.

Because of the rapid advances in data capture and data generation technologies (e.g., web-based data collection and Grid computing), the management of data over a complex computing and communication infrastructure is becoming a bottleneck to interactive visual data mining. Knowledge-based out-of-core algorithms were developed for managing very large datasets during visualization. Such algorithms can adapt their management strategies automatically based on data, resources and visualization requirements, hence removing the burden of complex data management from users.

Autonomic computing refers to computing systems which possess the capability of self-knowing and self-management. In addition to pursuing the integration of this technology with visualization systems, the group has also been actively developing autonomic capabilities, in industrial software systems, for managing large scale data transactions and complex job scheduling