Forest Fire Monitoring System Based on Image Analysis

Abstract

Large forest fires are very hazardous and often cause greater losses. However, due to the vast coverage of forests and the environmental complexity, monitoring which relies on sensor detection of direct contact almost cannot achieve its monitoring function. In this paper, the author proposes that long-range images will first be acquired and so do the images of a larger dynamic range of the observed region, then an embedded video monitoring system will be formed realizing a dynamic monitoring. With the techniques of dynamic image analysis and processing, dynamic image alarms and control signals can be obtained and such videos will be transmitted over the Internet in real-time. Experiments shows this system respond to fires with a fast speed of less than 20ms and the control of fire is highly accurate with a dynamic image area that is less than 1% of the background area.