Tracking - Illumination Neutralisation with application in tracking
Duration: 24 months (2010 - 2012)
Start: 15/9/2010
Project Website: https://graphics.cs.ucy.ac.cy/research/projects/illumination
Principal Investigator: Γιώργος Χρυσάνθου
Main Funding Source: Cyprus Research Promotion Foundation
Total Cost: 24,204
China has a vast human potential and has traditionally been supplying many of the best students and researchers at universities worldwide. However, the current effort is to develop this research potential locally and enhance it through external collaborations. Computer Graphics in particular, has seen a tremendous growth over the last decade in China. Since 1998 when Microsoft established a research center in Beijing, the graphics community has expanded rapidly with Chinese researchers becoming frequent contributors to top venues such as ACM SIGGRAPH, Eurographics, and IEEE Visualization. Following up from the visit of Prof B. Chen to Cyprus (2006), we recently visited FPA and other Graphics labs in China. We can indeed testify that there is great interest for international collaboration and sharing of know-how.
The objective of this proposal is to bring together the two main public academic institutions of Cyprus with a major research institution in China. The collaboration will be achieved through the implementation of a joint research project that lies at the intersection of the interests of the three parties. The aim of the research is to use advances in computer graphics (CG) techniques to help solve a classical computer vision (CV) problem: tracking. Traditionally CV approaches try to derive all the required information from the input images. This can lead to techniques that are unnecessarily compute-intensive and sensitive. Modern capture techniques can provide, in a fast way, information about the scene and incoming illumination. With this as input we will develop a novel approach for real-time illumination neutralization. The “neutralized” images will the be used as an input to the tracking module to increase its accuracy.
The task will require the development of several components making use of the complementary expertise of the three partners. The HO has expertise in CG techniques for illumination from the environment whereas the FPA are experts in scene capturing and CV. The proposed method will be addressing a very important problem in CG/CV but equally importantly it will form the beginning of a longer term collaboration since the partners have a number of different common interests in the wider area of 3D urban reconstruction and simulation.
In addition, it will give access to the Cypriot partners to a large pool of good student that could help fill in the gap. The largest current area of applications for CG & CV techniques is gaming. An area that all three partners have an interest. A second objective of this proposal is to investigate the future potential of computer game research and industry in Cyprus with a view to how this field has developed in China and the wider South-East Asia region.