基于递归双边滤波的快速图像去雾

Fast image dehazing based on recursive bilateral filtering

提出了一种基于递归双边滤波的单个图像去雾方法。结合场景辐射的边界条件,利用去雾模型估计大气耗散函数的初始值,采用递归双边滤波对其进行精化以获取高质量的清晰图像。根据当前帧视频图像若干相邻帧的信息来估计其全局大气光,利用递归双边滤波对视频图像的大气耗散函数进行跨帧滤波处理,将所提出的方法扩展应用于视频图像,获得时空连贯性的视频图像去雾结果。利用近邻传播聚类算法获取有雾图像的采样点以加速去雾计算。与现有方法相比,提出方法对含有大量边缘特征及复杂场景的有雾图像去雾效果更好。

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.