关键词: 图像去雾, 分割补偿, 景深估计图, 半衰减因子, B通道补偿

Abstract: 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.

Key words: Image fogging, segmentation compensation, depth of field estimation chart, half-decay factor, B channel compensation