Research on Fire Prediction Algorithm Based on Thermal Infrared Image

In order to realize the early prediction of fire, it is easy to make timely response to the potential fire accident and reduce the harm. In this paper, using a low-cost thermal infrared image sensor MLX90621 thermal image acquisition target area, first by Hue Saturation Value (HSV) color space transformation, temperature and chromaticity associated form two-dimensional temperature field image, then with the help of image threshold segmentation lock in high temperature area, at last, by calculating temperature field of exponential smoothing value high temperature area, and combining time series prediction model, to predict the temperature value for the future, to predict fire accident probability. The test results show that this algorithm can realize fast and reliable fire identification and prediction in a long distance, large range and all weather. It can be widely used in oil and gas, chemical industry, coal mine and other high-risk environments.