摘要
视频监控系统在日常生活中日益普及,为了实现监控视频的智能分析,利用背景差分法检测并提取关注的前景目标,再结合光流算法分析目标的运动方向和强度。为了提升视频分析效率,先利用帧差欧几里得距离法计算帧与帧之间的相似度,提取视频序列的关键帧,并提出了一种运算量小检测效果较好的基于图像灰度化的背景差分法进一步提取关键帧的前景目标,最后用光流法计算得到目标的光流矢量和强度信息,为监控视频的异常判断提供依据。
Video surveillance system has been increasingly popular in daily life.In order to realize the intelligent analysis of the monitoring video,the background difference method is used to detect and extract foreground targets and then the optical flow algorithm is used to analyze the targets ’ moving direction and intensity.To improve the efficiency of the video analysis,firstly,the frame difference Euclidean distance method is used to calculate the similarity between frames and to extract the key frames of video sequence.Then a background difference method based on greyscale image is proposed to extract the foreground targets of the key frames.This method has small computation and good detection.Finally,the optical flow method is used to calculate the optical flow vector and intensity of the targets,which provides the basis for the abnormal judgment of the monitoring video.
作者
张霞
贺正然
ZHANG Xia;HE Zhengran(School of Information Engineering,Putian University,Putian Fujian 351100,China;School of Information Engineering,Southeast University,Nanjing 210096,China)
出处
《电子器件》
CAS
北大核心
2019年第3期718-721,共4页
Chinese Journal of Electron Devices
基金
国家自然科学基金项目(61673108)
莆田市科技计划项目(2014G17)
关键词
灰度值
背景差分法
帧差欧几里得距离
光流法
gray value
background difference method
frame difference Euclidean distance
optical flow method
