An Embedded Low Power High Efficient Object Tracker for Surveillance Systems

This paper introduces a real-time surveillance application for object tracking with capabilities of functioning on 25 f/s mode, implemented in a high efficient, high performance and low power smart camera WiCa (wireless camera) developed by NXP Research. The proposed application deals with the problem of constantly changing environments as light change, swaying branches, rain and noise introduced by the camera with techniques of robust background modeling. This article also proposes a method to detect objects merging into the background model named "forgetting foreground" technique and also a new method called "multi-directional image mapping" used for object labeling and segmentation.