APP下载

基于邻域窗口滤波的图像去雾

2020-12-23李锋张朝霖

软件 2020年10期

李锋 张朝霖

摘  要: 户外摄影成像质量常常受雾霾等能见度低的天气影响,为了提高雾天的成像质量,提出了一种基于邻域窗口滤波——NWF(Neighborhood Window Filtering)的图像去雾算法。在大气散射模型下使用暗原色统计先验,大气透射率和暗通道有关,直接使用暗通道估计出的大气透射率结果粗糙,还原的图像边缘有光晕看起来不够真实。针对这一问题设计了一种简单高效的邻域窗口滤波器从八个邻域方向上计算切尾均值自适应选择出最佳的方向来提取精细的大气透射率图,该方法不仅能平滑尘雾浓度图,又能从根本上保留纹理信息。结合大气散射模型使用具有更多纹理边缘信息的的透射率图,最后可以还原出一幅更加真实的去雾图片。实验结果表明,该方法相比于传统的导向滤波及双边滤波的去雾方法更加简单,效果明显。

关键词: 去雾;邻域窗口滤波;切尾均值;暗通道;大气散射模型

中图分类号: TP391.41    文献标识码: A    DOI:10.3969/j.issn.1003-6970.2020.10.028

本文著录格式:李锋,张朝霖. 基于邻域窗口滤波的图像去雾[J]. 软件,2020,41(10):114119

【Abstract】: The quality of outdoor photography imaging is often affected by weather with low visibility such as haze. In order to improve the imaging quality of foggy days, this paper proposes an image defogging algorithm based on Neighborhood Window Filtering (NWF). This method is based on the atmospheric scattering model and the dark channel prior technique, and uses a neighborhood window filter with smooth texture preservation function to restore high-quality pictures. The feature of edge-preserving filtering using the domain window filter can extract more transmittance maps with texture edge information from the dark channel prior model. Combined with the physical model of atmospheric scattering, it can finally restore a high-quality defogging picture. The experimental results show that the method is simpler and more effective than traditional guided filtering and bilateral filtering defogging methods, and the effect is obvious.

【Key words】: Defogging; Neighborhood window filtering; Trimmed mean; Dark channel; Atmospheric Scattering model

0  引言

當空气中存在着许多悬浮物如尘埃、烟、水雾等时无论是人肉眼观察还是拍摄获取的图像,总是存在能见度和对比度下降的问题。这种降质退化的图像对我们后续的监控、识别、跟踪等实际应用场景带来了困难,因此对雾天成像去雾算法的研究是十分有意义和实用价值的。

针对雾天图像存在严重降质的现象,1999年,Srinivasa G. Narasimhan[1]等人通过建立数学模型,解释了雾天图像的成像过程以及雾天图像包括的各个要素。……

登录APP查看全文