To provide fire safety for high performance buildings, various types of fire/smoke detection systems are developed. Video fire detection is one of the important aspects in the development of fire detection system. It is particularly useful in large spaces with high headroom and buildings with cross ventilation design where traditional spot type smoke detection methods may not be effective. For the development of video fire detection system, spatial, spectral and temporal parameters are used to identify the fire source. One of the parameters captured by the video fire detection system is the flame height. With the information of flame height, real time heat release rate of fire can be estimated which is a very important parameter in determining the smoke generation rate and fire severity. Such information is very important in assisting evacuation and smoke control. In this study, experiments of pool fires with different pool diameters of 100mm, 200mm, 300mm and 400mm are conducted in the fire chamber of the laboratory in Department of Building Services Engineering, The Hong Kong Polytechnic University. The flame images, room temperatures and mass loss rates of the fuel are measured. The flame images are segmented using multi - threshold algorithm in a modified Otsu method and Rayleigh distribution analysis (modified segmentation algorithm). The algorithm use the optimum threshold values calculated to extract the pool fire images from a video sequence. After segmentation, flame height information can be obtained. In addition, other flame characteristics are also used for recognizing the flame region including flame color, flame light intensity, flame shape, and flicker frequency. Once the flame height is identified by the system, the heat release rate can be estimated using the equation developed by McCaffrey. The calculated heat release rates are then compared with measured heat release rate data. The results show that using flame height image for estimating real time heat release rate is promising.
展开▼
