Multi-camera based human tracking with non-Overlapping fields of view

This paper presents approach for an automated surveillance system which performs human detection and tracking across multiple non-overlapping cameras. At single camera level, motion based segmentation is achieved using an accurate optical flow estimation technique. Feature matching and region-based shape descriptors are used for object tracking and classification. The proposed approach extends these features across fields of view (FOV) of multiple non-overlapping cameras in order to obtain correct inter-camera correspondences. Performance evaluation of the system was done in different scenarios which verify that the proposed method can efficiently tackle the complexity of video motion tracking with high processing speed.