Influenced by the bad weather, such as haze, the images acquired from computer vision systems are seriously degraded, especially under driving scenario. The image features such as contrast and color will change, which makes intelligent transportation systems (e.g., auxiliary drive systems, automatic pilot systems) fail to extract image features, and even leads to traffic accidents. Therefore, it is significant to propose an effective and fast image dehazing method to improve the reliability of these systems. In this paper, we propose a novel fast image dehazing algorithm under driving scenario based on the dark channel prior. On the one hand, according to the image features of driving scenario, we simplify the process of estimating the atmospheric light. And on the other hand, in the process of refining the transmission map, we introduce the fast guided filter to our algorithm. Both of the two measures speed up the dehazing process obviously as well as keep a good dehazing effect. Demonstrated by the experiment results, our algorithm achieves a good dehazing effect and a fast speed, and meets the real-time requirement under driving scenario.