Detecting abnormal crowd behaviors based on the div-curl characteristics of flow fields.

Xiao-Han Chen,Jian-Huang Lai
DOI: https://doi.org/10.1016/j.patcog.2018.11.023
IF: 8
2019-01-01
Pattern Recognition
Abstract:•Explore the crowd state change detection problem for the dense crowd scenarios of wide-field of view in the public place.•Use a less computation required thermal diffusion processing and perspective transform processing to capture the macroscopic crowd motion information.•Model motion information by use of divergence and curl characteristics of the flow field.•Exploit a physical characteristic descriptor of crowd motion (PCM).•Calculate the conditional entropy of PCM distribution to perceiving the crowd changes.
What problem does this paper attempt to address?