skip to content

Multiview imaging

Supervisors: Prof. Laurence Dooley, Dr. Patrick Wong

Student: Mr. Mobolaji Akinola

This research project aims to address some of the exciting challenges arising from the development of NEXT GENERATION IMAGING technologies, which include multi-camera surveillance, 3DTVand free viewpoint TV. New paradigms are evolving to shift the traditional focus from more pixels, in terms of higher and super-definition imagery, to more views, with the corresponding change in emphasis from traditional pixel based to ray-based systems which afford multi-views of a scene. There are many research problems to be solved, especially with regard to efficient processing of the huge amounts of data generated by MVI. Multi-view coding (MVC) techniques seek to exploit both the spatial and temporal correlations existing not only within single-view videos, but also between multi-view sequences obtained from camera arrays. The use of cheap camera arrays to acquire image data from various sources and locations has also regenerated interest in shifting the processing paradigm from the traditional method of exploiting the spatial and temporal correlations at the encoder to the decoder due to new constraints of the use of cheap cameras with limited processing power and energy for image acquisition.

4 Views of a Dance Scene, courtesy of Tel-Aviv University VIA Laboratory.