Multi-sensor target detection and tracking system for sea ground borders surveillance

Abstract

Border safety is a critical part of national and European security. This paper presents a vision-based system for ground and maritime surveillance using fixed and moving PTZ cameras. This system is intended to be used as an early warning system by local authorities. For the ground surveillance scenario, we introduce a stable human tracker able to efficiently cope with the trade-off between model stability and adaptability. More specifically, we adopt probabilistic mixture models like the Gaussian Mixture Models (GMMs) which exploit geometric properties for background modelling. Then, we integrate iterative motion information methods, concerned by shape and time properties, to estimate image regions of high confidence for updating the background model. For the maritime surveillance scenario for ship detecting and tracking, the system incorporates a visual attention method exploiting low-level image features with an online adaptable neural network tracker. No assumptions about environmental or visual conditions are made. System performance was evaluated in real time for robustness compared to dynamically changing visual conditions with videos from cameras placed at a test area near Athens for the ground scenario and at Venetian port of Chania.