基于暗通道理论的雾天图像复原的快速算法

在雾天条件下捕获的视频或图像将会出现对比度下降、颜色偏移等严重的退化现象,这将极大的影响图像的主观视觉效果,大大降低其应用价值。传统的基于暗通道先验信息的全局最优化处理方法虽能获得较好的效果,但因其算法复杂而不具有实效性。本文将基于暗通道先验信息,利用双边滤波进行局部优化,从而获得保持边缘的暗通道图像,进一步利用该暗通道图像进行传输图像的估计,并最终复原场景信息。试验结果表明,此方法可较快的恢复场景信息并能有效保留场景的边缘信息。

A Fast Haze Removal Algorithm Using Dark Channel Prior

Images or videos captured under poor weather condition are characterized by poor visibility of scene, low vividness, loss of detail information, and color-shift phenomenon. Using Dark Channel Prior proposed by He , the haze removal algorithm based on the global optimization is computational complexity and quite time-consuming. In this paper, we exploit a new algorithm based on Gauss Bilateral Filter which can obtain an edge-preserving dark channel image. We further utilize this dark channel image to extract the estimation of medium transmission, and finally recover a haze free image from that. Experiments demonstrate that our algorithm can effectively remove haze from a foggy im age while keep edges sharp.