计算机工程

• 图形图像处理 • 上一篇    下一篇

基于多尺度图像融合的单幅图像去雾算法

潘磊1,2,郑益军2   

  1. (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)
  • 收稿日期:2015-07-20 出版日期:2016-08-15 发布日期:2016-08-15
  • 作者简介:潘磊(1982-),男,讲师、博士研究生,主研方向为图像处理、计算机视觉;郑益军,硕士研究生。
  • 基金项目:
    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)。

Single Image Dehazing Algorithm Based on Multi-scale Image Fusion

PAN Lei  1,2,ZHENG Yijun  2   

  1. (1.College of Computer Science,Civil Aviation Flight University of China,Guanghan,Sichuan 618307,China; 2.College of Computer Science,Sichuan University,Chengdu 610065,China)
  • Received:2015-07-20 Online:2016-08-15 Published:2016-08-15

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。

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

Abstract: To effectively improve the quality of the degraded image in foggy days,this paper proposes a dehazing 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.

Key words: single image dehazing, multi-scale, image fusion, image pyramid, multi-resolution fusion algorithm

中图分类号: