摘要
针对有雾天气下无人机航拍视觉系统的能见度低,航拍图像对比度和色彩保真度差等问题,基于暗原色先验规律以及雾图的物理模型提出了一种雾天降质图像去雾处理技术。从图像复原和增强两个角度出发,分别建立了户外图像全局去雾和对比度自适应调整的最优化模型,从而能够直接复原得到高质量的去除雾干扰的图像并且估算出雾的浓度。对一系列户外带雾图像的分组实验表明,该方法可以快速有效地提高带雾图像的对比度和色彩清晰度,获得满意的视觉效果。另外,该方法克服了Kai ming He方法处理时间过长的缺陷,平均处理时间仅为原方法的10%左右,显著缩短了运算时间,为在工程项目中实现图像的实时去雾处理提供了理论依据。
For aerial images with poor contrast and color fidelity due to foggy and hazy weathers,this paper proposes a technique of haze removal for aerial degraded images based on the dark-channel prior and the physical model to improve the visibility of vision system in an Unmanned Aerial Vehicle.From the viewpoints of image restoration and image enhancement,the optimized models of global haze removal and self-adapting contract extending are established,respectively.Using the method,a high quality haze-free image can be recovered and the thickness of the haze can be also established.The experimental results on a variety of outdoor haze images demonstrate that it can enhance the contrast and color definition of hazy degraded images fast and efficiently and can achieve satisfactory visual effects.Moreover,the method overcomes the Kaiming He＇s drawback of more time consuming,and the aver-age processing time is 10% that of the traditional method.It provides a theoretical reference for the real-time haze removal processing in engineering projects.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2011年第7期1659-1668,共10页
Optics and Precision Engineering
基金
国家863高技术研究发展计划资助项目(No.2008AA121803)
国家973重点基础研究发展规划资助项目(No.2009CB72400603B)
关键词
航拍图像
图像复原
图像增强
去雾
暗原色先验
aerial image
image restoration
image enhancement
haze removal
dark-channel prior
