电子学报 ›› 2017, Vol. 45 ›› Issue (5): 1218-1225.DOI: 10.3969/j.issn.0372-2112.2017.05.027

• 学术论文 • 上一篇    下一篇

基于先验知识与大气散射模型的图像增强算法

鞠铭烨, 张登银   

  1. 南京邮电大学物联网学院, 江苏南京 210003
  • 收稿日期:2015-08-05 修回日期:2015-11-09 出版日期:2017-05-25
    • 作者简介:
    • 鞠铭烨 男,1989年1月出生,江苏靖江人,博士生,研究方向模式识别与机器视觉.E-mail:2014070245@njupt.edu.cn;张登银 男,1964年11月出生,江苏靖江人,研究院、博士生导师.主要研究领域是下一代信息网络、信号与信息处理、网络信息安全.
    • 基金资助:
    • 国家自然科学基金 (No.61571241); 江苏省产学研前瞻性联合研究项目 (No.BY2014014); 江苏省高校自然科学研究重大项目 (No.15KJA510002)

Image Enhancement Based on Prior Knowledge and Atmospheric Scattering Model

JU Ming-ye, ZHANG Deng-yin   

  1. School of Internet of Things, Nanjing University of Posts and Telecommunications, Nanjing, Jiangsu 210003, China
  • Received:2015-08-05 Revised:2015-11-09 Online:2017-05-25 Published:2017-05-25

摘要:

针对现有图像增强算法大多不具备处理多种类型降质图像的能力,提出一种基于先验知识与大气散射模型的快速图像增强算法.首先,通过大量实验统计,提出一种新的图像先验—明亮通道先验,即高质量清晰图像中每个像素邻域都极有可能存在白点;随后,对散射模型所存在的缺陷加以改进,并结合明亮通道先验与黑色通道先验,推导出场景反射率的恢复公式;最后,针对黑色通道先验失效情况,提出一种基于可靠性预测的容错机制,以提高其适用范围.实验结果表明:本文算法不但可以有效的突出纹理细节,还具有一定的色调恢复功能,能够处理多种不同类型的降质图像.

关键词: 图像增强, 明亮通道先验, 黑色通道先验, 大气散射模型

Abstract:

Almost the existing image enhancement algorithms are not capable to deal with multi-type degenerated images.In this paper,a fast image enhancement algorithm is proposed based on prior knowledge and atmospheric scattering model.Firstly,an innovative image prior,which were called bright channel prior,were proposed through experimental statistics.It shows that the whites spots are more likely to exist in the neighborhood of each pixels in high-quality clear image.Then,the scattering model were improved and it derived the calculation formula of reflectance image by bright channel prior,dark channel prior and improved scattering model.Finally,we put forward a fault tolerance mechanism,which is based on reliability prediction,to handle the situation that the dark channel prior does not work.The experiment results shows that this algorithm can not only highlight the texture details well but also restore the tone.It is able to deal with multi-type degenerated images.

Key words: image enhancement, bright channel prior, dark channel prior, atmospheric scattering model

中图分类号: