Fuzzy Feature-Based Upper Body Tracking with IP PTZ Camera Control

摘要

In this paper, we propose a fuzzy-feature based method for online upper body tracking using an IP PTZ camera. Because the camera uses a built-in web server, camera control entails camera response time and network delays, and thus, the frame rate is irregular and in general low (2-7 fps). It detects at every frame, candidate targets by extracting motion, a sampling method, and appearance. The target is detected among samples with a fuzzy classifier. Results show that our system has a good target detection precision (> 85%), low track fragmentation, and the target is almost always localized within 1/6th of the image diagonal from the image center.