Pedestrian's Abnormal Behaviour Detection Based on Machine Vision. Issue 1 (1st April 2022)

• Record Type: Journal Article
• Authors: Zhao, Rongyong; Jia, Ping; Yanwang, ; Zhu, Wenjie; Li, Cuiling; Ma, Yunlong
• Is Part Of: Journal of physics. Volume 2224:Issue 1(2022)
• Publication Date: 2022-04-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2224/1/012019
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: It is significant to automatically analyse video sequences for timely and accurate detection of abnormal pedestrian behaviour by using computer vision and analysis technology in public places. To realize the real-time monitoring of abnormal behaviour, this study uses a hybrid behaviour detection model based on the colour characteristics of the pixels and the gradient characteristics of the pixels mathematically. According to the hybrid behaviour detection model, the criterion of abnormal behaviour of the same pedestrian based on two frames of video images is proposed. This is conducive to realizing early awareness of emergencies caused by abnormal behaviours and taking early warning measures, reducing casualties, and maintaining public safety and social stability.