摘要
对4种光学影像上薄雾的处理方法进行了比较研究,目的是为有效开展海岛区域测绘提供高质量的影像。据海岛区域影像的特点,重点改进了基于暗原色先验知识的去雾方法,包括暗通道的快速优化和大气光的分块求取两个方面。该方法首先计算影像的暗通道并进行优化,然后求取影像的大气光;再根据图像恢复模型得到去雾的影像,并与同态滤波法、平稳小波变换法和点扩展卷积法进行了比较。实验表明4种方法均可在一定程度上去除影像上的薄雾,对基于暗原色先验知识的改进方法效果最佳。
In order to provide qualified remote sensing images,four methods for island area images dehazing were studied.According to the characteristics of island images,the method based on dark channel prior was emphatically improved by fast optimizing dark channel and finding air-light via dividing an image into a few blocks.This method is implemented as follows： firstly,the dark channel for an image is calculated and optimized;secondly,air light value of the images is obtained;last,image is restored through the imaging model.This improved method was compared with the other three methods including homomorphic filtering,stationary wavelet transform and point spread inversion.The results showed that the improved method based on dark channel prior is well done haze reduction in four methods.
出处
《测绘科学技术学报》
北大核心
2010年第4期266-269,274,共5页
Journal of Geomatics Science and Technology
基金
国家863计划资助项目(2009AA121403)
关键词
大气光
同态滤波
薄雾影像
平稳小波变换
暗原色先验知识
air-light
homomorphic filtering
haze image
stationary wavelet transform
dark channel prior
