基于B通道补偿的景深估计的图像去雾算法

Research on Fast Defogging Algorithm Based on Depth Evaluation of B Channel Compensation

为提高图像去雾效果,兼顾初始细节与亮度,提出一种基于B通道补偿的景深估计快速去雾算法。首先分析大气散射模型,根据B通道与雾浓度关系,构造B通道景深估计模型,计算景深,利用R、G通道分量绝对差补偿B通道景深;为防止近景过度补偿和远景漏补,利用图像的B通道分量来构造分割补偿模型,对远、近景的像素灰度值进行补偿。且通过设定半衰减因子,修正景深图,形成景深评估图;最后利用最小滤波和引导滤波优化景深图,实现图像去雾效果。实验结果显示:与当前图像去雾技术相比,所提算法具有更好的去雾效果,更好地保持了图像细节与亮度。

In order to improve the effect of image defogging and take into account the initial details and brightness, a fast defogging algorithm based on B-channel compensation for depth of field estimation is proposed. Firstly, the atmospheric scattering model is analyzed. According to the relationship between B-channel and fog concentration, the depth of field estimation model of B-channel is constructed, and the absolute difference of R and G-channel components is used to compensate the depth of field of B-channel. In order to prevent the over-compensation of near-field and the missed compensation of long-range, the B-channel component of image is used to construct the segmentation compensation model to compensate the gray value of far-range and near-range pixels. The depth-of-field assessment map is formed by setting half-attenuation factor and modifying depth-of-field map. Finally, minimum filter and guide filter are used to optimize depth-of-field map to achieve image defogging effect. The experimental results show that, compared with the current image defogging technology, the proposed algorithm has better defogging effect and better preservation of image details and brightness.