基于引导滤波的单幅图像自适应去雾算法  

Single Image Adaptive Defogging Algorithm Based on Guidance Filtering

作　　者：杨燕[1] 白海平[1] 王帆[1] 

机构地区：[1]兰州交通大学电子与信息工程学院,兰州730070

出　　处：《计算机工程》2016年第1期265-271,共7页Computer Engineering

基　　金：国家自然科学基金资助项目(61561030);甘肃省科技厅自然科学基金资助项目(1310RJZA050);甘肃省财政厅基本科研业务费基金资助项目(214138);兰州交通大学研究生校内教改立项基金资助项目(160012)

摘　　要：在雾、霾等天气条件下,由于空气中悬浮粒子的大气散射作用导致获得的户外图像严重降质。针对该问题,提出一种单幅图像自适应去雾算法。在分析大气散耗函数特性的基础上,利用引导滤波估计大气散耗函数,间接求得透射率,通过自适应选取阈值调整明亮区域透射率,改善复原图像中的假彩色,并根据大气散射模型恢复图像。实验结果表明,与传统引导滤波算法和取固定阈值的算法相比,该算法获得的去雾图像近景更清晰,明亮区域恢复更自然,同时提高了运算速度。Outdoor images are often severely degraded due to atmosphere scattering of suspended particles on the condition of haze and foggy weather.In order to solve the problem,this paper proposes a single image adaptive defogging algorithm.Transmission is estimated indirectly by the method of accurate atmospheric veil through guided filtering on the basis of analysis of characteristic for atmospheric veil.Transmission of bright areas can be adjusted utilizing adaptive threshold value which obviously reforms the false colors in defogging images.At last,the images are recovered in the way of taking advantage of atmosphere scattering model.Experimental results show that compared with traditional guidance filtering algorithm and selecting stationary threshold value algorithm,the obtained defogging images have clearer nearby view and more natural bright areas,and the computational speed is improved.

关 键 词：去雾算法 引导滤波 阈值 复杂度 大气散射模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]