Real-time multi-view vision systems using WSNs

  • Authors:
  • Paolo Pagano;Francesco Piga;Yao Liang

  • Affiliations:
  • ReTiS laboratory, Scuola Superiore Sant'Anna, Pisa (I);ReTiS laboratory, Scuola Superiore Sant'Anna, Pisa (I);Indiana University Purdue University Indianapolis, Indianapolis (IN)

  • Venue:
  • Proceedings of the 2009 ACM symposium on Applied Computing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multi-View Vision systems are collaborative distributed applications devoted to image processing. Applications based on Multi-View Vision systems can recognize shapes, track moving targets, etc. to activate alarms, or to propogate information through the network. These systems require Quality of Service from the network and real-time support at the Operating System level. Wireless Sensor Networks (WSNs) were born for monitoring applications providing best effort services in non-critical environments. Nonetheless the recent evolution of hardware platforms, and the existence of real-time kernels as well as network stacks supporting real-time traffic let us envisage the deployment of WSNs to new domains like that of Multi-View Vision. Hereby we discuss the feasibility of an object detection system based on vision and deployed through a WSN. For such a purpose, we implemented in the Real-Time Network Simulator (RTNS) a working model for image detection and in-network processing. Referring to a simple star-shaped network scenario, we analyze the system performances from a real-time perspective.