一种联合雾线和凸优化的单幅图像去雾算法

为解决传统去雾算法容易在天空区域出现光晕效应和复原后的图像颜色过饱和等问题,提出了一种联合雾线和凸优化的单幅图像去雾算法。该算法使用雾线先验来估计大气光值,并通过离散小波变换构建了一个降维的子带雾图模型,进一步将双线性耦合项和大气光传输分布作为线性优化变量进行凸优化求解来得到透射率,最后通过大气散射模型恢复出无雾图像。实验结果表明该算法在大多数情况下恢复的图像清晰自然,与其他几种常用的图像去雾算法的客观对比,也证实了该算法的可行性和有效性。

A single image dehazing algorithm combined with haze line and convex optimization

In order to solve the problems of halo effect and color oversaturation in the sky region,a single image dehazing algorithm is proposed in this paper.The algorithm bases on haze line prior to estimate atmospheric optical value,and through the discrete wavelet transform construct a subband haze image model of dimension reduction,further double linear coupling term atmospheric optical transmission and distribution as a convex optimization to solve linear optimization variables for transmittance.At last,the haze free image is recovered through the atmospheric scattering model.The experimental results show that the image recovered by this algorithm is clear and natural in most cases,and the objective comparison with other commonly used image dehazing algorithms also proves the feasibility and effectiveness of this algorithm.