PhD Defense: "Contour-based Object Tracking using Simultaneous Registration and Segmentation"

Pratim Ghosh

May 24th (Tuesday), 10:00am
Harold Frank Hall (HFH), Rm 4164

Tracking objects in image sequences is a fundamental problem in computer vision. Robust tracking is critical in many vision applications, including surveillance and security systems, medical image analysis, and entertainment industry. However, the tracking problem is extremely challenging due to the high degree of uncertainty associated with the observed data. In recent years, considerable research effort has been devoted to developing solutions in controlled experimental settings.

In this thesis, we present several tracking algorithms applicable to a wide range of challenging scenarios. In particular, we focus on a specific sub-class of tracking methods, referred to as contour based object tracking, which simultaneously tracks and segments an object of interest in a given image sequence. The presented algorithms are developed based on the theory of simultaneous registration and segmentation. The intuition of adopting this theory comes from the fact that temporal statistics of the tracked object exhibit strong correlation in image sequences. Robustness is further achieved by incorporating various prior information available from specific application domain. The results have been demonstrated on a wide variety of data; and have been presented at leading conference venues and journals.

About Pratim Ghosh:

Pratim Ghosh was born in West Bengal, India in 1981. He received his B.E. degree in Electrical Engineering from Jadavpur University, Kolkata in 2004 and M.E. degree in System Science and Automation (with distinction) from Indian Institute of Science, Bangalore in 2006. He has interned at Janelia Farm, Howard Hughes Medical Institute in the summer of 2009. He is currently a PhD. student in the Department of Electrical and Computer Engineering at the University of California, Santa Barbara. His research interests span the areas of computer vision and pattern recognition with particular emphasis on variational methods for image segmentation and tracking.

Hosted by: Professor B.S. Manjunath