Structured dictionary learning for abnormal event detection in crowded scenes

作者：

Highlights：

• Dominating event patterns are discovered as reference events.

• Relationships among video events are described with a smoothness regularization.

• SR framework is constructed with reference events and smoothness regularization.

• The identification of abnormal events gets easier.

摘要

•Dominating event patterns are discovered as reference events.•Relationships among video events are described with a smoothness regularization.•SR framework is constructed with reference events and smoothness regularization.•The identification of abnormal events gets easier.

论文关键词：Video surveillance,Abnormal event detection,Dictionary learning,Sparse representation,Reference event

论文评审过程：Received 19 November 2015, Revised 2 May 2017, Accepted 2 August 2017, Available online 3 August 2017, Version of Record 18 September 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.08.001