Accurate detection of moving objects in the video is an important precursor to a wide spectrum of computer vision related applications. Background modeling and subtraction is very a popular and fast approach for moving object detection. However background modeling for real-time applications in complex outdoor environments is a challenging task due to a number of issues.
In this seminar some key approaches from the literature for background modeling will be presented along with their pros and cons. In addition the challenges and issues in the background modeling domain in general will be highlighted. The focus of the talk will be on real-time approaches for complex background models for videos taken with a stationary camera.
In light of these approaches I will present some of our recent experimental results to achieve better foreground object detection in dynamic environments using our proposed "adaptive Mixture model learning" and "enhanced CodeBook model" for background modeling. At the end I will present future trends and open questions in the background modeling domain.
Last modified: Tuesday, 26-Jul-2011 10:14:34 NZST
This page is maintained by the seminar list administrator.