Abnormal global and local event detection in compressive sensing domain

Tian Wang,Meina Qiao,Jie Chen,Hichem Snoussi,Wenjia Zhang,Chuanyun Wang

doi:10.1063/1.5022779

Abstract

Abnormal event detection, also known as anomaly detection, is one challenging task in security video surveillance. It is important to develop effective and robust movement representation models for global and local abnormal event detection to fight against factors such as occlusion and illumination change. In this paper, a new algorithm is proposed. It can locate the abnormal events on one frame, and detect the global abnormal frame. The proposed algorithm employs a sparse measurement matrix designed to represent the movement feature based on optical flow efficiently. Then, the abnormal detection mission is constructed as a one-class classification task via merely learning from the training normal samples. Experiments demonstrate that our algorithm performs well on the benchmark abnormal detection datasets against state-of-the-art methods.

Highlights

Though having been extensively researched, abnormal event detection remains a challenging and important problem in the surveillance video research
While detection methods are largely dominated by one-class classification techniques, various feature representation algorithms have been proposed by researchers in the literature
After learning from the training normal samples by one-class SVM, the support vector and hyper-plane are obtained for anomaly detection

Summary

Introduction

Though having been extensively researched, abnormal event detection remains a challenging and important problem in the surveillance video research. Due to the occlusion and appearance change, tracking method remains a challenging problem.[6] Feature descriptors, such as co-occurrence matrix,[7] pixel change history,[8] mixture of dynamic texture,[9] histograms of oriented swarms with histograms of gradients,[10] convolutional neural network,[11] were proposed for event analysis. These methods relied on the semantic segmentation performance to a certain extent.

Methods

Results

Conclusion