摘要
由于传统方法在视频监控图像异常行为识别应用中效果不佳,漏识率较高,且识别用时较长,文章为此提出基于神经网络的视频监控图像异常行为识别方法。该方法首先分割处理视频监控图像,然后应用视频帧图像阵列对目标图像进行预处理并提取图像行为特征,最后建立神经网络模型,利用模型计算图像行为特征卷积,完成异常行为识别。实验证明,该方法的漏识率为0.56%,具有较好的应用前景。
Because the traditional methods are not effective in the application of abnormal behavior recognition in video surveillance images,the missing rate is high,and the recognition time is long,a method of abnormal behavior recognition in video surveillance images based on neural network is proposed.Firstly,the video surveillance image is segmented and processed,then the target image is preprocessed by using the video frame image array,and the image behavior characteristics are extracted.Finally,the neural network model is established,and the convolution of the image behavior characteristics is calculated by using the model to complete the abnormal behavior identification.Experiments show that the missing rate of this method is 0.56%,and it has a good application prospect.
作者
刘云萍
LIU Yunping(Department of Computer Engineering,Taiyuan Institute of Technology,Taiyuan 030008,China)
出处
《数字通信世界》
2023年第7期71-73,共3页
Digital Communication World
基金
山西省教育厅项目(J20221105)
院级教改项目(JG202009)。
关键词
神经网络
视频监控图像
异常行为
漏识率
neural network
video surveillance image
abnormal behavior
misrecognition rate