方紫娟, 谢正祥, 吕霞付, 冯鹏. 一种改进的基于暗通道先验知识的去雾算法[J]. 微电子学与计算机, 2014, 31(5): 74-78.
引用本文: 方紫娟, 谢正祥, 吕霞付, 冯鹏. 一种改进的基于暗通道先验知识的去雾算法[J]. 微电子学与计算机, 2014, 31(5): 74-78.
FANG Zi-juan, XIE Zheng-xiang, LU: Xia-fu, FENG Peng. An Improved Haze Removal Algorithm Based on Dark Channel Prior[J]. Microelectronics & Computer, 2014, 31(5): 74-78.
Citation: FANG Zi-juan, XIE Zheng-xiang, LU: Xia-fu, FENG Peng. An Improved Haze Removal Algorithm Based on Dark Channel Prior[J]. Microelectronics & Computer, 2014, 31(5): 74-78.

一种改进的基于暗通道先验知识的去雾算法

An Improved Haze Removal Algorithm Based on Dark Channel Prior

  • 摘要: 针对上述问题首先分析了基于暗通道先验知识的去雾算法存在的缺陷与不足,然后通过预处理把原图像转换成标准图像,接着以引导滤波器代替软抠图来细化信息图以提高算法运行的速度,并调整信息图来改善原算法在处理明亮区域的失真和光晕现象.通过实验证明,改进的暗通道先验知识去雾算法比原算法在速度上提高了84.2%,并避免了对明亮区域的过度处理而产生的失真现象,增强了对色彩单一的有雾图像的处理效果.

     

    Abstract: In order to deal with the fog image more effectively,an improved method based on haze removal using dark channel prior is presented in this paper after analyzing its defects and shortcomings.First of all,the original image is converted into a standard image through standardization transformation,and then we improve speed by replacing the soft matting with guided filter to refine transmission map and adjust transmittance in bright region to avoid the distortion and halo phenomenon.The experiment results show that the improved algorithm in speed increased by 84.2% than the original algorithm,and refrained from distortion which caused by excessive processing in bright region,and enhanced processing effects on color single foggy images.

     

/

返回文章
返回