航拍图像去雾优化算法研究

河北工业大学电子信息工程学院,天津 300401

图像处理; 航拍图像去雾; 暗通道; 四叉树; 快速导向滤波; 多尺度Retinex; 直方图去雾

Optimization algorithm of aerial image dehazing
LIU Cuixiang, ZHANG Sha, WANG Baozhu, and YUAN Xiangwei

School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, P.R.China

image processing; aerial image dehazing; dark channel; quadtree; fast guided filter; multiscale Retinex; histogram defogging

DOI: 10.3724/SP.J.1249.2018.05487

备注

为优化航拍图像的去雾算法,提出基于暗通道规律的快速去雾、四叉树子空间划分层次搜索算法和快速导向滤波整合处理的综合去雾优化方法,发现相比原单一算法,新算法的复杂度被大幅降低,同时提高了图像清晰度.通过研究不同算法的图像处理结果发现,单独使用某一算法只能优化图像的某一方面,整体效果不明显,而采用这3种优化算法综合处理后的航拍有雾图像,清晰度至少提高11.57%,峰值信噪比至少增加1.88%,综合优化算法耗时比单一优化算法至少缩短了37.49%,不仅去雾效果好且实时性高.分别对雨天和雪天情况下的雾气图像进行实验分析,结果证明,该综合算法对雨天和雪天的有雾航拍图像同样有效.分别采用多尺度Retinex算法、直方图去雾算法和本研究提出的综合去雾算法对航拍有雾图像进行去雾处理,本算法清晰度高,更适合航拍图像.

In order to study the optimization algorithm of aerial image dehazing, we propose a combined defogging method based on three optimization algorithms, including a rapid defog based on dark channel, the use of a hierarchical search method based on quadtree subspace to optimize the acquisition of atmospheric light, and optimizing the transmission based on the fast-directed filtering. We find that our algorithm can greatly reduce the complexity compared to any single original algorithm and improve the image clarity. By studying the image processing of different algorithms, we find that the use of only one of the three algorithms can only optimize the characteristics of a certain aspect of image and the effect is not obvious. But using these three optimization algorithms to simultaneously process aerial foggy images, the resolution is improved by at least 11.57%, the peak signal-to-noise ratio is increased by at least 1.88%, and the time is at least 37.49% shorter than that of a single optimization algorithm, that is, the defogging effect is good and the real-time performance is high. At the same time, to verify the feasibility and universality of the research algorithm, we analyze the fog images in rainy and snowy days experimentally. We also compare the results of our combined defogging algorithm to those of the multi-scale Retinex algorithm and histogram dehazing algorithm to verify the applicability of the algorithm. The experimental results show that the proposed algorithm has the wide applicability.

·