低照度视频监控图像降噪算法设计与实现  

Design and Implementation of Noise Reduction Algorithm for Low-light Video Surveillance Image

作　　者：刘磊磊[1] 蒋荣欣[1] 

机构地区：[1]浙江大学浙江省网络多媒体技术研究重点实验室,杭州310027

出　　处：《计算机工程》2014年第3期266-269,共4页Computer Engineering

基　　金：国家"863"计划基金资助项目"相控阵三维声学摄像声纳信号处理系统"(2010AA09Z104)

摘　　要：针对低照度环境下视频监控图像噪点较多的问题,提出一种基于运动检测的低照度视频监控图像降噪算法。在研究低照度视频监控图像噪声特点的基础上,通过一种阈值运动检测算法将图像帧划分成8×8的运动像素宏块和静止像素宏块,对运动像素宏块采用改进的维纳滤波算法进行降噪,对静止像素宏块采用数学形态学和中值滤波相结合的算法进行降噪。实验数据显示,该算法总体时间复杂度接近O(n),使用该算法降噪后的图像的PSNR值和DV,BV值均高于经典降噪算法,证明了该算法在降低时间复杂度的同时,能有效降低图像噪声,并较好地保持图像的解析度。For strong noise of low-light video surveillance image, a new image noise reduction algorithm based on motion detection is proposed. The property of the noise of low-light video surveillance image is studied and the image is divided into 8×8 motion pixels blocks and still pixels blocks by a kind of threshold motion detection algorithm. An improved Wiener filter is designed and implemented for noise reduction of motion pixels blocks. The compact algorithm of mathematical morphology and median filtering for noise reduction of still pixels blocks is designed. Experimental results show that the time complexity of the algorithm is about O（n） and the value of PSNR and DV/BV of image after noise reduction is higher than other algorithms. This proves the time complexity of the image noise reduction algorithm is low, while the image noises are well reduced and there is little loss in the resolution of image.

关 键 词：低照度 图像降噪 运动检测 数学形态学 中值滤波 维纳滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]