基于多尺度图像融合的单幅图像去雾算法  

Single Image Dehazing Algorithm Based on Multi-scale Image Fusion

作　　者：潘磊[1,2] 郑益军 

机构地区：[1]中国民用航空飞行学院计算机学院,四川广汉618307 [2]四川大学计算机学院,成都610065

出　　处：《计算机工程》2016年第8期255-260,共6页Computer Engineering

基　　金：中国民航总局应用开发科技基金资助项目(MHRD20140212);四川省教育厅科研基金资助项目(16ZB0032);中国民用航空飞行学院面上基金资助项目(J2012-40)

摘　　要：为了有效增强雾天降质图像的质量,基于多尺度图像融合技术,提出一种单幅图像去雾算法。将图像转换到梯度域,采用透射率计算各尺度值并进行梯度域增强及重建,避免因尺度值选择不当造成增强后的图像色彩失真。根据各尺度增强图像的对比度和饱和度计算其相应的权重,再利用多分辨率融合算法对经过不同尺度增强后的图像进行融合,从而得到最终的清晰图像。实验结果表明,该算法在主观视觉效果和客观评价指标上均优于现有图像去雾算法,且耗时更短。To effectively improve the quality of the degraded image in foggy days, this paper proposes a delaazmg algorithm for single image based on multi-scale image fusion. After the image is converted to gradient domain, each scale value is calculated by means of transmissivity and the gradient domain is strengthened and restructured to avoid color distortion after enhancement caused by improperly chosen scale values. The corresponding weight is calculated according to the contrast and saturation of enhanced image of each scale. After enhanced by different scale, the image is fused by means of multi-resolution fusion algorithm. Finally, the clear image is obtained. Experimental results show that this algorithm is superior to current dehazing algorithms in terms of subjective visual effect and objective evaluation index,and it is more time-efficient.

关 键 词：单幅图像去雾 多尺度 图像融合 图像金字塔 多分辨率融合算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]