为了解决遥感图像中存在雾气引起图像可见度下降的问题, 提出一种基于颜色线先验和暗通道先验的遥感图像去雾算法。针对颜色线先验对部分场景不适用的情况, 提出一种基于最小二乘的透过率优化方法, 通过计算先验可信度权重来加权融合颜色线先验和暗通道先验估计的透过率图, 以获得更为准确的透过率分布。为了避免halo效应, 在优化方程中加入梯度正则化约束。实验结果表明, 本方法对透过率估计准确, 重建结果清晰度提升明显。另外, 该方法改善了客观评价指标, 其灰度平均梯度(GMG)和信息熵(IE)平均值最高, 取得了最佳的重建效果。

A color-line and dark channel based dehazing algorithm was proposed to solve the problem of image contrast decreasing caused by haze in remote sensing images. In view of the deficiencies in the color-line a priori, a least squares based transmittance optimization method was proposed. In order to obtain a more accurate transmittance distribution, the transmissivity maps estimated by the color-line prior and the dark channel prior are fused by weight which is calculated by their credibility. A gradient regularization constraint was added to the optimization equation to avoid the halo effect. The experimental results show that the proposed method is accurate in estimating transmittance and the clarity of the reconstruction results is significantly improved. In addition, the objoctive evaluation index is improved as its Gray Mean Grads(GMG) and Information Entropy(IE) show the highest average value, which indicates that the proposed method has achieve the best reconstruction result among the state-of -art methods.