摘要
作为计算机视觉的重要分支,异常行为识别与检测技术已在社会安防、人工智能、交通管控等领域获得了广泛应用。针对不同应用场景特点,选择适当的特征提取及异常行为识别与检测方法,进而保证实时预警准确率,保障社会公众安全,在实际应用中至关重要。基于此,文章对基于视频的人群异常行为识别与检测方法进行综述,首先,对人体异常行为中的目标检测算法作一介绍;其次对特征提取方法加以总结,特征提取方法的选取及提取特征的准确与否直接影响后续判别结果;之后,从异常行为识别和异常行为检测两个方面的主流算法进行归纳,并总结常用异常行为检测方法相关性能参数;最后,对该领域未来研究方向提出了展望。
As an important branch of computer vision, abnormal behavior recognition and detection technology has been widely used in social security, artificial intelligence, traffic control and other fields. According to the characteristics of different application scenarios, it is very important to select appropriate feature extraction and abnormal behavior recognition and detection methods, so as to ensure the accuracy of real-time early warning and ensure the safety of the public. Based on this, this paper reviews the methods of crowd abnormal behavior recognition and detection based on video. Firstly, it introduces the target detection algorithm in human abnormal behavior. Secondly, it summarizes the feature extraction methods. The selection of feature extraction methods and the accuracy of the extracted features directly affect the subsequent discrimination results. Then, from two aspects: abnormal behavior recognition and abnormal behavior detection Finally, the future research direction in this field is proposed.
作者
周彤彤
彭月平
郑璐
蒋镕圻
Zhou Tongtong;Peng Yueping;Zheng Lu;Jiang Rongqi(Engineering University of PAP,Xian 710086,China)
出处
《无线互联科技》
2022年第6期93-95,127,共4页
Wireless Internet Technology
关键词
群异常行为
特征提取
异常行为识别
crowd abnormal behavior
feature extraction
abnormal behavior recognition