基于暗原色融合和维纳滤波的单幅图像去雾

Single Image Dehazing Based on Dark Channel Fusion and Wiener Filtering

针对在雾霾等恶劣天气下捕获的户外场景图像对比度降低、颜色失真等问题,对基于暗原色先验的去雾方法进行改进,应用小波变换将块暗原色和点暗原色进行融合后,得到新的透射率估计,并利用自适应维纳滤波细化透射率.同时提出了四分加权法重新估计大气光,使得大气光更具鲁棒性.实验结果表明,本文方法不仅能有效恢复清晰的无雾图像,而且能够大幅提升运行速度,便于实时应用.

Haze is one of the major factors that cause color distortion and contrast loss of the outdoor image. To re-duce these effects, an improved dehazing method based on dark channel prior is proposed in this paper. A new trans-mission is estimated after the fusion of patch-based dark channel prior and pixel-based dark channel prior using wave-let transform, and adaptive wiener filter is introduced to further refine the transmission estimation. Simultaneously, in order to make the result more robust, we propose a method called weighted quadtree subdivision to estimate atmos-pheric light. Comparative experimental results demonstrate that the proposed method is effective not only in restoring the clear images, but in speeding up the computation, which is appropriate for real time application.