基于时间—空间滤波器的视频异常抖动过滤  

Video abnormal jitter filtering based on time-space filter

作　　者：龙年 刘智惠 LONG Nian;LIU Zhi-hui(Hubei University of Technology Engineering and Technology College,Hubei Wuhan 430000 China)

机构地区：[1]湖北工业大学工程技术学院,湖北武汉430000

出　　处：《计算机仿真》2022年第3期210-214,共5页Computer Simulation

摘　　要：视频异常抖动消除过程中进行视频异常抖动检测,容易出现视频异常抖动消除效果不显著,峰值信噪比下降,平均编码时间上升的问题,为此提出基于局部特征约束的视频异常抖动消除,提取视频中的运动特征符,通过时间—空间滤波器对特征描述符进行平滑处理,获取对应区域的平滑估计。分别计算训练和测试视频中各个区域局部K-最近邻距离,获取两种视频的总体分值,对总体分值进行排序,确定视频异常抖动。同时估计相机运动路径,引入视频编码技术进行视频异常抖动消除。仿真实验结果表明,所提方法能够有效减少平均编码时间,增加峰值信噪比,有效消除视频中的异常抖动。The traditional video abnormal jitter detection has many defects,such as poor jitter elimination effect,low peak signal-to-noise ratio and long average coding time.Therefore,a video abnormal jitter elimination method based on local feature constraints is proposed.The motion feature in the video was extracted,and the feature descriptor was smoothed by a time-space filter in order to obtain a smooth estimation of the corresponding region.The local K-nearest neighbor distance of each region in the training and test video was calculated respectively,obtaining the overall score of both.The overall score was sorted to determine the abnormal jitter of the video.Concurrently,the camera motion path was estimated,and video coding technology was also applied to eliminate the abnormal jitter of video.The simulation results show that this method has reduced average coding time,increased peak signal-to-noise ratio and excellent effect of abnormal jitter elimination.

关 键 词：局部特征约束 视频异常 抖动消除 异常检测 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]