摘要
文章提出了一种基于暗原色先验的图像去雾新方法,此方法将中值滤波处理后的有雾图像的灰度图或灰度图的反转图作为优化透射率;根据归纳出的两条准则划分出天空区域,进而计算得到大气光值;最后借助有雾图像形成模型获得最终的去雾图像。实验表明,与同类方法相比,本方法在主观效果和客观指标上均能够获得较高质量的清晰图像;而且克服了同类算法耗时的缺点,缩短了算法的运行时间。
This paper proposes a novel method for image dehazing based on dark channel prior. With median filter, this method can treat the foggy image＇s gray-scale image or its reversal gray-scale image as optimized transmittivity, divide the sky area by two inductive criteria, then calculate out the atmospheric optical value, and finally obtain the final dehazed image with the help of the foggy image formation model. Experiment shows that compared with some similar methods, our method can get sharp image with high quality, both on the subjective effect and the objective indicators, and overcome the defect of time-consuming, improving the running time of algorithm.
出处
《电子技术（上海）》
2015年第1期1-4,共4页
Electronic Technology
关键词
暗原色先验
灰度图
灰度图的反转图
图像去雾
dark channel prior
gray-scale image
inverse gray-scale image
image dehazing
作者简介
董林娜，山东师范大学物理与电子科学学院，硕士电话：15253169528电子信箱：tclzakdg@163．com联系地址：山东省济南市历下区文化东路88号山东师范大学物理与电子科学学院（250014）
