基于大气散射模型的单幅图像快速去雾  

Fast single image fog removal based on atmospheric scattering model

作　　者：孙伟[1] 李大健[2] 刘宏娟[2] 贾伟[2] 

机构地区：[1]西安电子科技大学机电工程学院,陕西西安710071 [2]西北工业大学第365研究所,陕西西安710065

出　　处：《光学精密工程》2013年第4期1040-1046,共7页Optics and Precision Engineering

基　　金：国家自然科学基金青年基金资助项目(No.61201290);中央高校基础科研业务费专项资金资助项目(No.K50511040008)

摘　　要：根据大气散射物理模型及光学反射成像模型,总结并分析了影响单幅图像去雾效果的3大因素,以实现对雾霾图像的快速去雾。基于光学原理,解释了暗影通道现象,从新的角度推导出了大气散射模型中各参数的求法。利用灰度开运算去除白色目标的干扰获得精确的环境光亮度,基于快速联合双边带滤波精确计算了大气散射函数,最后由光学反射模型计算了场景目标的反射率并有效截断至[0,1]区间。本方法可以消除天空及环境光线的影响,能真实复原场景的色彩和清晰度。仿真结果表明,对分辨率为576×768的图像处理时间仅为0.517s,且视觉效果和客观指标比现有算法均有不同程度的提高。与现有图像去雾算法相比,本文提出的参数计算方法提高了运算速度、场景适应能力和复原效果。Based on the physical model of atmospheric scattering and an optical reflectance imaging model, three major {actors influencing the fog removal for a single image were discussed in detail. The dark channel phenomenon was explained by the optical model, and the method to solve the parameters of atmospheric scattering model was rigorously derived from a new view. The gray-scale opening oper- ation was used to eliminate the interference from a while object to obtain the global atmospheric light and the fast joint bilateral filtering technique was proposed to greatly improve the speed and accuracy of atmospheric scattering function solving. Finally, the scene albedo was recovered by inverting this model. Experiments show that the method can remove effectively the effect of lights from sky and en- vironments and can recover the color and definition of original scenes. The simulation results indicate that the processing time for an image of 576 × 768 spends only by 1.7 s. As compared with the exist- ing algorithm, obtained results on a variety of outdoor foggy images demonstrate that the proposed method achieves good restoration for contrast and color deity, and improves image visibility greatly.

关 键 词：光学成像模型 大气散射物理模型 快速联合双边滤波 暗影通道 图像去雾 

分 类 号：TP751.2[自动化与计算机技术—检测技术与自动化装置]