摘要
为解决传统视频监控系统中异常行为检测与识别方法存在检测效率低、工作时间长的问题,提出了一种新的视频监控系统中异常行为检测与识别方法。该方法首先通过视频图像噪声过滤、图像灰度矫正、二值化处理、图像边缘检测4个步骤,完成图像预处理;然后在明确图像异常目标特征的基础上,对运动异常目标图像的关键帧进行检测与数据解剖,完成视频监控系统异常行为检测;最后通过自适应算法对视频图像规律加以分析,利用计算机的视觉检测随场景环境变化原理,识别视频监控系统的异常行为。为检测方法效果,设置了对比实验,实验结果表明,新方法能够在短时间内精准地检测出异常行为,工作能力强。
Aiming at low detection efficiency and long working time in the abnormal behavior detection and identification methods in traditional video surveillance systems,a new abnormal behavior detection and identification method in video surveillance systems is proposed.The four steps such as noise filtering,gray scale correction,binarization and image edge detection realize the image preprocessing,clarify the characteristics of abnormal target images,detect key frames of moving abnormal target images and analyze data,and complete abnormal behavior detection of video surveillance systems.The algorithm analyzes the laws of video images,and uses the principle of computer vision detection to change with the scene environment and identify abnormal behaviors of video surveillance systems.A comparative experiment results show that the new method can accurately detect abnormal behavior in a short time and have a strong working ability.
作者
董莹荷
胡国胜
Dong Yinghe;Hu Guosheng(School of Communication and Information Engineering, ShanghaiTechnical Institute of Electronics & Information, Shanghai, 201411, China)
出处
《机械设计与制造工程》
2020年第3期66-70,共5页
Machine Design and Manufacturing Engineering
关键词
视频监控系统
异常行为
识别
关键帧
目标检测
video monitoring system
abnormal behavior
identification
key frames
target detection