Abstract
The talk will address two computer vision challenges related to visual surveillance: person detection and active camera control. While a number of researchers have addressed the problem of detecting isolated people, segmenting groups of people is still an open problem. In the first part of the talk a novel approach to this problem will be presented which effectively integrates feature grouping and model based segmentation into one consistent framework. The algorithm is based on partitioning a given set of image features using a likelihood function that is parameterized on the shape and location of potential individuals in the scene. Using a variant of the EM formulation, maximum likelihood estimates of both the model parameters and the grouping are obtained simultaneously.

The second part of the talk will focus on active camera control, in particular how to model and detect salient events in the scene. A hierarchical background model will be presented that systematically treats static and stationary dynamic regions of the scene as background. The talk will illustrate on how this can be achieve using a moving pan tilt zoom camera.

Brief CV
Jens Rittscher joint the Visualization and Computer Vision Group at GE Global Research as a staff scientist in 2001. His research focuses on algorithm development for the automatic interpretation of video and time lapse imaging. Recently he participated in the VACE programme sponsored by ARDA and worked on projects for GE Security, Lockheed Martin, and GE Healthcare. He holds a diploma in Mathematics from Bonn University (Germany) and a doctorate degree in Engineering Science from Oxford University (UK). His thesis entitled 'Classification of Human Motion' was supervised by Prof. Andrew Blake. He holds a Marie Curie Fellowship awarded by the European Union and has published more than 20 peer reviewed publications in international journals and conferences and hold two U.S patents.