Abstract: In the event of a fire in a building, the internal CCTV system has poor monitoring images under the influence of hot smoke. A multi - scale Retinex (GL - MSR) fire defogging method combining Gaussian pyramid and log lookup table is proposed to improve the monitoring effect in complex fire situations. A fire image enhancement model based on GL - MSR was constructed, and the Gaussian pyramid and log lookup table method were used to reduce the image processing time. The experimental results show that the image processing time based on GL - MSR is 1/26 of the traditional MSR method. The smoke image processing effect is better in case of emergency lighting than infrared lighting. In comparison, the processed smoke image of GL - MSR is clearer, which can effectively improve the monitoring effect of the fire scene monitoring image.

Key words: fire smoke, image defogging'multi - scale Retinex, GL- MSR