一种雨景视频图像的复原方法  

A restoration algorithm for rain video image

作　　者：朱志超[1] 李斌[1] 方帅[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2011年第7期1011-1014,共4页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(60705015)

摘　　要：文章分析了视频中雨的时空特性和色度特性,并基于这2个特性,提出了一种简单而有效的利用序列图片检测和去除视频中雨的方法。首先优化了K均值的时间复杂度,然后依据雨的时空特性,运用改进的K均值方法对各个像素进行聚类操作;依据雨的色度特性准确区分出雨区,从而成功地完成了检测;利用统计特性对背景像素值进行估计,代替雨的像素值。为了得到更加清晰的复原结果,融合了检测之前的预处理和去雨后的非雨区滤波处理,形成了一套完整的雨景图像复原方法。实验结果表明,该方法是简单而且有效的。In this paper,the spatio-temporal property and the chromatic property of rain in the video are analyzed.Based on the two properties,a simple and effective algorithm is proposed to detect and remove the rain in the video by using sequential image.Firstly the time complexity of K-means is optimized by the algorithm.Then each pixel is classified by the optimized K-means according to the spatio-temporal property of rain and the rain parts are identified accurately according to the chromatic property of rain for the rain detection.For the rain removal,the color of streak is replaced by the proportion-blending color of the rain＇s color and the background＇s color based on the statistic characteristics.And for achieving a clearer restoration result,the original images are preprocessed before the detection and filtering is carried out for the no-rain parts after rain removal.Experimental results prove that the proposed algorithm is simple and effective.

关 键 词：雨的检测 雨的去除 图像复原 色度特性 K均值 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]