26 / 36
Abstract
: Security and monitoring systems are more and more demanding in terms
of quality, reliability and flexibility especially those dedicated to video surveillance.
The recent advances in electronic Computer Aided Design (CAD) tools have led to much
advanced hardware devices especially for multimedia and wireless applications. This has
resulted in increased deployment worldwide of sensor networks for visual/video
surveillance and security purposes which have gained much maturity due to the availability
of cost effective distributed sensor nodes. However, despite the tremendous progress
already made towards the development of efficient security systems, the existing solutions
have limitations especially in complex and cluttered environments such as the environment
in a busy soccer stadium or high traffic roads/highways. These difficulties could be
alleviated by using multi-sensor and multi-modal surveillance systems by exploiting the
redundancy and diversity of data provided by the acquisition system.
This project aims to develop a versatile platform for a security monitoring system
incorporating advanced techniques for multisensor signal pre- and post-processing, multi-
modal data and information fusion, and intelligent sensor connectivity and secure wireless
communications. The overall platform will be tested for video surveillance systems for
public security. In such systems, reliability is a key feature that is affected by signal
distortions due to the presence of artifacts resulting from technical limitations at different
stages of the communication process. We propose to use multidimensional and multi-scale
signal processing techniques in order to localize and mitigate the artifacts introduced by
environment and system limitations. We will develop biologically-inspired approaches to
fuse information collected from different sensors to exploit diversity and redundancy for
improving the efficiency of detection, recognition and tracking tasks. The developed
techniques will be tested on real scenarios corresponding to very low contrast and noisy
data acquired by our multi sensor platform.
Team:
Professor Azeddine Beghdad
Professor Ahmed Bouridane
Dr Noor Al Maadeed
Dr Somaya Al Maadeed
A Biologically-inspired Multi-sensor and Multi-modal
System for Public Security
Department Research
26