fire eye targets at the implementation of an early warning system for the in-time-detection of fires. In a highly wooded area flames of an upcoming fire are hidden by the intense tree population. In this case the fire detection has to be carried out via the heat deployment.
The main target was the development of a computer-visions-algorithm acting as the detector of fire.
fire eye consists of a ground-based surveillance station which is fitted with a colour camera including a motorized zoom objective and a thermal camera. These cameras are mounted tiltable and rotatable and with a search angle of 360°. The surrounding area of the station can be monitored completely within a certain radius. The great advantage of thermal cameras is that with the right calibration they can optically display fire and - according to the type - can operate over long distances and completely independent from the actual weather situation.
The surveillance sensors autonomously and permanently work off the developed algorithms and communicate steadily with the control station of the operator. All surveillance stations can be operated and administrated by the control station. In case during the picture analysis a predefined value triggers an alarm, the operator is informed with an alarm message (SMS, E-Mail, acoustic signal). Furthermore the analysed values including the pictures are transferred to the control station. The operator can access the surveillance station in real-time and confirm the accuracy of the alarm.
The multi-dimensional sensor data (spectral and spatial) pass through a multilevel processing:
• Colour analysis
• Texture analysis
• Motion analysis
• Thermal data analysis
Only the combination of these results allows a reliable detection of fire with an acceptable `False-Positive`-rate close to 100%.
So an area of a several square kilometres can be covered. To be able to guarantee a comprehensive surveillance of affected regions multiple sensors have to operate in a network.
In comparison to already existing systems the essential improvement for the product fire eye chosen approach is the focus upon the evaluation/interpretation of different information delivered by the sensors (colour/thermal cameras), which run through logical algorithms. Due to the modular design stand-alone installations (application area: timber industry, storage areas etc.) or cross-linked (network) installations (WLAN) in swaths of land are possible.
• Scanning the landscape within a short period of time and therefore extremly
quick in alerting
• Scanned area 300km2 from one station
• Accuracy nearly 100%
• Online system
• Cost efficient compared to existing systems
• Easy handling
• Minimum labour costs
• Further area of application: industrial areas, timber industry, lumber mills,
storage areas etc.