A single image dehazing method based on recursive bilateral filtering is proposed. Combining with the boundary conditions of scene radiance, the dehazing model is applied to estimate the initial value of the atmospheric veil, and recursive bilateral filtering is applied to refine the initial value in order to obtain the high-quality clear image. Based on several neighboring frames information around current frame of the video-image, its globe atmosphere light is estimated. The atmospheric veil is filtered across each frame of the video-image with recursive bilateral filtering. The proposed method is extended to video-image dehazing, and the video-image dehazing result with temporal coherence is obtained. The affinity propagation algorithm is applied to obtain sampling points of haze images for accelerating the dehazing processing. Compared with the existing methods, the proposed method can get better dehazing results for the haze images with many edge features and complex scenes.