Abstract: Coal mine intelligent video monitoring system in the coal mine are widely used. However,
due to the harsh down hole conditions, video surveillance frequently encountered accompanied by a
variety of random noise of fog image. In view of the serious degradation and poor visual effect of the
monitoring image, a fast image defog method based on the twice bilateral filtering (FDA-DBA) is
proposed. First getting the environment brightness by quadtree, and then using the bilateral filter
towards minimum channel map to obtain coarse scattering map and do further optimization to get
high precise transmittance map, secondly using the tolerance mechanism to correct the transmission
to get high precise transmission map and resolve color distortion in the bright area after dehazing,
finally restore the fog degraded image based on the atmospheric scattering model. Experiments show
that the algorithm in this paper can accurately recover the scene color and clarity, and get a real clear
no-fog image, and has higher accuracy and robustness, what’s more, the relationship between the time
complexity of this algorithm and the number of image pixels is a linear correlation, which makes the
algorithm in this paper is suitable for the environment of coalmine intelligent video surveillance.

Key words: fog and dust images restoring, twice bilateral filter, quadtree, bilateral filter, tolerance
mechanism