| 融合暗通道滤波和空间金字塔的图像去雾算法 |
| |
| 引用本文： | 宋佳怡,谢维信,王鑫.融合暗通道滤波和空间金字塔的图像去雾算法[J].信号处理,2019,35(5):816-824. |
| |
| 作者姓名： | 宋佳怡 谢维信 王鑫 |
| |
| 作者单位： | 深圳大学信息工程学院 |
| |
| 基金项目： | 国家自然科学基金重点项目:分数阶Fourier变换应用基础理论与关键技术及在海洋目标探测中的应用（61331021）
|
| |
| 摘 要： | 针对红外摄像机在有雾天气下红外图像降质的问题，本文提出一种融合暗通道滤波和空间金字塔的图像去雾算法（Dark channel filtering and spatial pyramid algorithm，后文简称DCF-SP）用于海上红外图像的去雾。首先，对海上红外图像进行自适应的海天分割，再将天空区域均分为四个子空间，定位灰度均值最大的子空间，重复迭代上述过程，将三级分割后的最亮子块中邻域像素亮度最大的值视为大气光估计值。接着，对清晰和带雾红外图像进行最小值滤波，得到清晰红外图像暗通道值趋于零的先验知识，最后估计出暗通道透射率，实现红外图像去雾。实验结果表明，对比现有去雾算法，DCF-SP算法鲁棒性更高运行时间更短，且更能保留图像目标的边缘信息，EPI和SSIM指标分别达到0.95和0.9425。
|
| 关 键 词： | 海上红外图像 图像去雾 大气光估计 暗通道先验 |
| 收稿时间： | 2018-12-31 |
Fusion of dark channel filtering and spatial pyramid image defogging algorithm |
| |
| Affiliation: | College of Information Engineering, Shenzhen University |
| |
| Abstract: | For the problem of poor infrared image quality in foggy weather, this paper proposes a dark channel filtering and spatial pyramid algorithm(DCF-SP) for defogging maritime infrared image.Firstly, the sea-sky region is segregated by the adaptive algorithm in the image.And then,the sky region of image is divided into four subspaces to locate the subspace with the largest average gray value.The maximum brightness of neighborhood pixels in the brightest sub-block is regarded as the atmospheric light estimate after three iterations.The minimum filtering of clear and foggy infrared images is carried out to obtain the prior knowledge that the dark channel value of clear infrared images tends to zero,finally,estimate the dark channel transmittance, and realize the haze removal of infrared images.Compared with the existing defogging algorithm, experimental results show that, the DCF-SP algorithm has higher robustness and shorter running time, and can retain the edge information of image targets more effectively. EPI and SSIM metric achieve 0.95 and 0.9425 respectively. |
| |
| Keywords: | |
|
| 点击此处可从《信号处理》浏览原始摘要信息 |
|
点击此处可从《信号处理》下载全文 |
|
