A classification method based on optical flow for violence detection

作者：

Highlights：

• The methodology has been focused on the intelligent video surveillance Systems.

• A new feature descriptor named HOMO for violence detection.

• HOMO has the ability to detect violence in both crowded and non-crowded scenes.

• HOMO outperforms baseline descriptors in non-crowded scenes.

• The performance of HOMO is better than OViF in crowded scenes.

摘要

•The methodology has been focused on the intelligent video surveillance Systems.•A new feature descriptor named HOMO for violence detection.•HOMO has the ability to detect violence in both crowded and non-crowded scenes.•HOMO outperforms baseline descriptors in non-crowded scenes.•The performance of HOMO is better than OViF in crowded scenes.

论文关键词：Violence detection,Optical flow,Adaptive thresholding,SVM classifier,Global thresholding

论文评审过程：Received 5 October 2018, Revised 6 February 2019, Accepted 24 February 2019, Available online 26 February 2019, Version of Record 12 March 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.02.032