Integrated Metric Learning Based Multiple Object Tracking Method under Occlusion in Substations. Issue 5 (August 2020)

• Record Type: Journal Article
• Authors: Zhang, Pengfei; Chen, Zhongyang; Zhang, Xinyue; Yang, Zhongguang; Shi, Wenbin
• Is Part Of: IOP conference series. Volume 558:Issue 5(2020)
• Publication Date: 2020-08
• Subjects: Earth sciences -- Periodicals; Environmental sciences -- Congresses; Environmental sciences -- Periodicals; 550.5
• DOI: 10.1088/1755-1315/558/5/052067
• Languages: English
• ISSNs: 1755-1307

Abstract

Abstract: Nowadays most of background subtraction based algorithms lack the robustness to handle tracking multiple objects with specific situations such as heavy occlusion in intelligent video surveillance system. This paper proposed an integrated metric learning based multiple object tracking method in smart substations. First tracks personnel obtaining bounding box, then integrates obtained mahalanobis distance and cosine distance of personnel. When occlusion occurs, proposed method compares the integrated metric vale with threshold so as to track personnel in different frames. Experiments in the smart substation confirm that our method can track multiple personnel under occlusion effectively and reliably.