Pre-Warning System for Sudden Events in Public Based on Pedestrian Behavior's Understanding from Surveillance Video

DOI

10.2991/cnct-16.2017.76How to use a DOI?

Keywords

K-medoids, Wavelet Neural Network, Surveillance Video Processing.

Abstract

Surveillance video records the situation of pedestrians in public. If some sudden event such as a robbery or someone gravely ill, guards or managers can get this by watching and understanding pedestrians behaviors from surveillance video in time. In this paper, we propose a system to warning the guard that there are something abnormal in public by analyzing people action from the video. This system contains three subsystem. Firstly, we adapt a wavelet neural network (WNN) for predicting the number of pedestrians. It sends alarm when the amount of people become excessively large. Secondly, we record the direction of every pedestrians and if most people go to the same direction instead of go to all the direction randomly, such situation should be treated with caution. Thirdly, the k-medoids clustering algorithm is used to calculate cluster center and cluster size of current pedestrian for describing distribution situation and give an alarm if the distribution becoming exceptional. Finally, we use a pedestrian database to exam our system.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Yuan-zhi LIANG
PY  - 2016/12
DA  - 2016/12
TI  - Pre-Warning System for Sudden Events in Public Based on Pedestrian Behavior's Understanding from Surveillance Video
BT  - Proceedings of the International Conference on Computer Networks and Communication Technology (CNCT 2016)
PB  - Atlantis Press
SP  - 556
EP  - 562
SN  - 2352-538X
UR  - https://doi.org/10.2991/cnct-16.2017.76
DO  - 10.2991/cnct-16.2017.76
ID  - LIANG2016/12
ER  -