基于暗通道先验的多尺度融合去雾算法
A Multi-scale Fusion Defogging Algorithm Based on Dark Channel Priors
贺文靖 1张俊生 2景宁 1王明泉 1成向北1
作者信息
- 1. 中北大学信息与通信工程学院 太原 030051
- 2. 太原工业学院电子工程系 太原 030008
- 折叠
摘要
针对传统暗通道先验(DCP)算法用于图像去雾时存在对比度下降、色调偏暗和过度曝光的问题,论文提出一种基于暗通道先验的多尺度融合去雾算法.首先,该算法使用图像增强暗部细节后采用阈值分割法将类似白色背景区域进行分离;然后对大面积留白背景和其他区域分别做最亮通道取反去雾处理和导向暗通道去雾处理,将去雾处理后的两部分图像进行初始融合,有效地改善了暗通道先验在大尺度高亮区域的不适用性;最后,使用拉普拉斯金字塔的多曝光融合方法,将上述去雾处理后的两幅图像、初始融合图像以及对有雾图像做导向暗通道处理后的图像进行多尺度像素级融合,获得最终去雾效果图.仿真结果表明,该算法能够有效消除传统DCP算法在处理大范围、高亮度天空区域时产生的失真现象,改善目标边缘细节,恢复图像场景的真实色彩.
Abstract
At the problems of contrast reduction,tone darkening and overexposure when the traditional dark channel prior (DCP)algorithm is used for image defogging,a multi-scale fusion defogging algorithm based on dark channel prior is proposed in this paper. Firstly,the algorithm uses the image enhancement to enhance the dark part details and then uses the threshold segmenta⁃ tion method to separate the similar white background areas. Then,for large white background and other areas,the brightest channel inverse defogging processing and the dark channel guided defogging processing are respectively carried out. The two parts of the im⁃ age after defogging are initially fused,which effectively improves the unapplicability of dark channel priors in large scale highlight⁃ ed areas. Finally,the Laplacian Pyramid multi-exposure fusion method is used to conduct multi-scale pixel-level fusion of the above two images after defogging,the initial fusion image and the image with fog after guiding dark channel processing,to obtain the final defogging effect. The simulation results show that the proposed algorithm can effectively eliminate the distortion caused by the traditional DCP algorithm when processing the large range and high brightness sky area,improve the target edge details,and re⁃ store the real color of the image scene.
关键词
图像增强/图像去雾/暗通道先验/曝光融合Key words
image enhancement/image defogging/dark channel prior/exposure fusion引用本文复制引用
基金项目
国家自然科学基金(61171177)
国家自然科学基金(62105302)
出版年
2024
国家科技期刊平台
NSTL
万方数据