Searching for humans lost in vast stretches of ocean has always been a difficult task. In this paper, a range of machine vision approaches are investigated as candidate tools to mitigate the risk of human fatigue and complacency after long hours performing these kind of search tasks. Our two-phased approach utilises point target detection followed by temporal tracking of these targets. Four different point target detection techniques and two tracking techniques are evaluated. We also evaluate the use of different colour spaces for target detection. This paper has a particular focus on Hidden Markov Model based tracking techniques, which seem best able to incorporate a priori knowledge about the maritime search problem, to improve detection performance.
Evaluation of Maritime Vision Techniques for Aerial Search of Humans in Maritime Environments
2008-12-01
320347 byte
Conference paper
Electronic Resource
English
Online Contents | 2002
Online Contents | 2002
Online Contents | 2000
Online Contents | 1998
Online Contents | 2003