Abstract: In order to improve ship navigation safety in fog,a real-time haze removal algorithm was proposed for marine surveillance video. For establishing fog degraded physical model,an inequality scaling theory was preliminarily used to estimate transmission and the global atmosphere for every frame of marine surveillance video. Then the haze-removal surveillance video was recovered based on the physical model. Finally,according to the marine surveillance video with large sky and sea areas,the parameters of the algorithm were estimated based on the factors of navigation target details and image contrast. The experiment results show that the algorithm can remove the haze in real-time speed,recover the marine targets' details and improve the image contrast. So,it proves that this algorithm can well solve the problem of real-time haze removal for marine surveillance video.