Removal of rain in video and images is an important issue that interests both digital image processing and computer vision areas, and it has been extensively used in many applications such as video surveillance, automatic transmission, and military reconnaissance. The essential goal of rain removal in video and images is how to detect raindrops or streaks and remove them, and restore the details of the scenes. In order to make more researchers have a more comprehensive understanding in rain removal techniques in video and images, a detailed introduction was reviewed. On the basis of extensive literature research, the basic ideas, methods, characteristics and effect of various kinds of rain removal techniques in videos and images are summarized and compared. According to the implementation ways and disposal methods, rain removal techniques in videos and images are classified into four types, they are imaging parametric-based method, time domain-based method, frequency domain-based method, and sparse domain-based method. At the same time, the analysis of recent years in time domain, frequency domain and sparse domain to the latest developments in the rain is focused. With the improvement of computer hardware performance and the deep research of raindrop characteristics, the method research and application conditions for rain removal in videos and images have made great progress. However, due to the interference in rainy dynamic complexity and/or the environment with complex background, rain removal techniques in videos and images are still far from perfect, there are still many open problems that need to be solved.%视频图像去雨是数字图像处理和计算机视觉领域共同关注的重要问题,在视频监控、智能交通乃至军事侦察等领域有着广泛的应用.为了使国内外同行对视频图像去雨技术的发展历程和研究现状有一个较为全面的了解,对其进行系统综述.在广泛文献调研的基础上,对各种视频图像去雨技术的原理、特点以及处理效果进行归纳总结和分类比较.根据处理方式和实现途径,将视频图像去雨技术区分为基于成像参数的去雨方法、基于时域的去雨方法、基于频域的去雨方法和基于稀疏域的去雨方法四大类.重点介绍了这些方法的基本思想和基本原理,分析比较了各自的优点与不足.随着计算机硬件性能提高以及对雨滴特性研究的不断深入,视频图像去雨技术在方法研究和应用条件上都取得了很大的进展,但由于雨天的动态复杂性以及环境中复杂背景的干扰,导致视频图像去雨技术研究还不完善,仍然存在很多亟待解决的问题.
展开▼
