Based on the nighttime traffic monitoring images, the observation visibility data of nearby automatic stations are chosen as the criteria for whether the image is foggy. The effects of foggy and fogfree environments on light sources, and their imaging are analyzed; and the corresponding features are extracted. A recognition model is established to identify fogs by the method of the support vector machine. Results show that, due to the increase of air extinction coefficient in the presence of fogs, the light from LED electronic display or cars and their reflection in the monitoring image weaken. The image gradient, texture and other parameters are obviously different in foggy and clear weather. The critical success index of the model evaluation result is 0.74, and the false alarm rate is 0.18, which meet the needs of fog monitoring at night.