Outdoor Fire Detection Based on Color and Motion Characteristics

With due attention to industry deployment and extension of urban zones, early warning systems have critical role in giving emergency response to unexpectedevents. Video-base fire detection is a low cost and effective method for this purpose. Most of available fire detection methods only use color information in detection process thatis inaccurate. This paper intends to increase the accuracy of fire detection in video sequences using motion detection andcombination of two classifiers. Movement of pixels and their color in the YCbCr space are considered for detection. Using this combined method, false alarms due to movements of ordinary objects with fire-like color, are greatly reduced in comparison with other color based fire detection systems

کلیدواژه ها:

fire detection ، motion detection ، YCbCr color space ، Gaussian Mixture Models ، support vector machines