Author(s)

A. Losso, L. Corgnati, S. Bertoldo, M. Allegretti, R. Notarpietro & G. Perona

Abstract

Forest fires in our society cause a lot of damage, in particular regarding the economic and environmental landscape. In order to monitor a large portion of territory automatically, with a good cost/performances trade-off, it is necessary to develop new early warning systems. We propose a ground-based system with modular architecture, equipped with low cost commercial sensor. The idea is to develop the software able to manage forest fire monitoring. The technique is based on static and dynamic analysis of chromatic changes between images, tailored for our case of study in a large scale monitoring of vegetation and using different sensors to reduce or eliminate the false alarm rate. Concerning the image geo-referencing tool, the present work describes an innovative projective geo-referencing algorithm able to geo-reference complex orography regions using fixed ground station images. Besides, it does not need the collection of Ground Control Points, which is a very hard task in complex orography environments. In order to make a user oriented product and to help the operator during extinguishing activities, a decision support tool has been developed as well. This work presents the results of one year monitoring campaign conducted in cooperation with the Civil Protection Offices in Sanremo (IM), Italy. Keywords: forest fires, smoke detection, geo-referencing, decision support, image processing, low cost commercial sensors.

Keywords

forest fires, smoke detection, geo-referencing, decision support, image processing, low cost commercial sensors.