High-speed object tracking across multiple networked cameras

This paper presents a networked vision system for tracking objects moving across multiple camera views at high speed. Frame acquisition in each camera is triggered by a personal computer (PC), which also computes the target position and sends data to another PC via an Ethernet link. The data includes the tracked target position and the timestamp when the frame acquisition was triggered. The timescale on which the timestamps are based is precisely synchronized between the PCs by using the precision time protocol (PTP). Therefore, locally acquired data in each PC and data received via Ethernet can be precisely aligned in time. A 5 mm-diameter object moving across two cameras at 4000 mm/s at a distance of 300 mm from the cameras was successfully tracked by the system with an average frame rate of more than 720 frames-per-second.

[1]  Masatoshi Ishikawa,et al.  A Networked High-Speed Vision System for 1,000-FPS Visual Feature Communication , 2007, 2007 First ACM/IEEE International Conference on Distributed Smart Cameras.

[2]  Akio Namiki,et al.  1-ms sensory-motor fusion system , 2000 .

[3]  Masatoshi Ishikawa,et al.  Vision chip architecture using general-purpose processing elements for 1 ms vision system , 1997, Proceedings Fourth IEEE International Workshop on Computer Architecture for Machine Perception. CAMP'97.

[4]  M. Branicky,et al.  Design Considerations for Software Only Implementations of the IEEE 1588 Precision Time Protocol , 2005 .

[5]  Devavrat Shah,et al.  Network Coding Meets TCP: Theory and Implementation , 2011, Proceedings of the IEEE.