|
|||||
|
|
| 基于暗原色先验与Retinex理论的去雾算法 | |
| 引用本文： | 舒婷,邓波,陈炳权,刘耀峰. 基于暗原色先验与Retinex理论的去雾算法[J]. 吉首大学学报(自然科学版), 2014, 35(2): 52-56. DOI: 10.3969/j.issn.1007-2985.2014.02.012 |
| 作者姓名： | 舒婷 邓波 陈炳权 刘耀峰 |
| 作者单位： | （吉首大学物理与机电工程学院，湖南 吉首 416000） |
| 基金项目： | 湖南省自然科学基金资助项目（12GJ6055）；湖南省科技厅科技计划项目（2010FJ4107）；吉首大学科研资助项目（13JDY014） |
| 摘 要： | 为提高雾霾图像的清晰化程度，采用暗原色先验方法与多尺度Retinex(MSR)算法相结合的方式，提出基于暗原色先验与Retinex理论的去雾算法.首先，采用暗原色先验方法对协雾霾图像进行物理意义上的去雾，雾霾图像细节得到增强；其次，将处理后的雾霾图像作为MSR算法的原始输入图像，雾霾图像对比度与平均亮度增大.仿真实验表明，该算法处理的雾霾图像对比度得到增大，图像平均亮度适中，总体效果较好，可在一定程度上提高雾霾图像的清晰度. |
| 关 键 词： | Retinex算法 图像增强与复原 直方图 标准差 熵 |
Defogging Algorithm Based on the Prior Dark-Channel and Theory of Retinex |
|
| SHU Ting,DENG Bo,CHEN Bing-Quan,LIU Yao-Feng. Defogging Algorithm Based on the Prior Dark-Channel and Theory of Retinex[J]. Journal of Jishou University(Natural Science Edition), 2014, 35(2): 52-56. DOI: 10.3969/j.issn.1007-2985.2014.02.012 | |
| Authors: | SHU Ting DENG Bo CHEN Bing-Quan LIU Yao-Feng |
| Affiliation: | (College of Physics and Mechanical & Electrical Engineering，Jishou University，Jishou 416000,Hunan China) |
| Abstract: | To improve the sharpness of the hazy image，utilizing the prior dark-channel method and the multiscale Retinex (MSR) algorithm，a defogging algorithm based on the prior dark-channel and the theory of Retinex is proposed in this paper.First of all，the prior dark-channel method is applied to dehaze the original image in physically，and the detail of image is thus enhanced.Then，the image obtained is treated as the originated input of the MSR algorithm.After that the contrast and average brightness of the final image are improved.Simulation shows that the contrast of hazy image is increased，the average brightness of image is moderate，and the overall effect is good.Furthermore，the clarity of hazy image can be improved in some extent. |
| Keywords: | Retinex algorithm image enhancement and restoration histogram standard deviation entropy |
| 本文献已被 CNKI 维普 等数据库收录！ | |
| 点击此处可从《吉首大学学报(自然科学版)》浏览原始摘要信息 | |
| 点击此处可从《吉首大学学报(自然科学版)》下载全文 | |