Early Fire Detection in Coalmine Based on Video Processing

Fire is one of the most serious catastrophic disasters in the coalmine. The early fire detection can help to avoid a disastrous fire. The existing temperature-sensed and smoke-sensed method may respond slowly to early fire, if it is far away from the sensors, or the setting values of sensor are unsuitable. Therefore, the detection method based on image and video processing is adopted to overcome the drawbacks of traditional fire detection method for coalmine. The paper analyses the status in the fire detection technology, and designs a structure to detect the early fire in coalmine. Firstly, image which comprises the potential fire region the potential fire region is detected by using frame differencing of monitor video, and denoised by median filter. Secondly, flame region is extracted by color information. Finally, Bayes classifier is employed to recognize fire combined with the dynamic features. The method can greatly improve accuracy of early fire prediction in coalmine, comparing with the traditional detection method.