The School of Electrical, Computer, and Telecommunications Engineering (SECTE) cordially invites you to our research seminar.
Topic: Optimizing the arrangement of multiple cameras for effective capturing of visual information
Presenter: Prof. Farzad Safaei (SECTE, Faculty of EIS, University of Wollongong)
Date/time: Thursday April 21, 2016, 13:30-14:30
Venue: 35-G20, Main Campus, UOW
Brief Biography: Farzad Safaei graduated from the University of Western Australia with the degree of Bachelor of Engineering (Electronics, HONSI) and obtained his PhD in Telecommunications Engineering from Monash University in 1998. Currently, he is the Professor of Telecommunications Engineering at the University of Wollongong. His research interests include multimedia and wireless communications.
Multiple cameras are frequently used in many applications, such as visual surveillance and broadcasting of sports events. The scenario in which a single camera attached to a robot moves to different locations to scan a scene can also be considered as a multiple camera system. This is referred to as visual simultaneous localisation and mapping (SLAM). Traditionally, the main purpose of using multiple cameras is to cover the scene, and the placement of cameras is decided by human experience. The recent development in immersive virtual reality and visual SLAM requires reconstructing the scene, i.e. generating 3D geometric models of the scene objects with surface texture from the images captured by the cameras. In this case, the theory of camera geometry alone cannot evaluate whether a given camera placement is sufficient and appropriate for reconstruction, because the camera geometry does not take the scene into consideration. In this talk, I will present the result of our research in the development of mathematical models and algorithms for optimising the arrangement of cameras for the purpose of reconstructing the scene. In particular, our focus is on two criteria. One is the quality of depth estimation, since accurate depth estimation is a key to quality 3D modelling of the scene. The other is the quality of arbitrary view synthesis, since view synthesis is essential to many immersive systems including the future free-viewpoint video (FVV). We have developed two mathematical models, called correspondence field of cameras and the effective sampling density that can be used to optimise the camera arrangement.?
Eis_academics mailing list Eis_academics@mailman.uow.edu.au http://mailinglists.uow.edu.au/mailman/listinfo/eis_academics