An architecture to identify violence in video surveillance system using ViF and LBP

Abstract

There are different varieties of Surveillance cameras used but it is still a challenge to detect violence. So the aim is to design a violence detection system which detects violence and generates an alert so that help will be available instantly. Researchers are prognosticating that the evolution of video surveillance technology will lead to a great demand for intelligent violence detection system. In coming years also, these technological advancement will continue by improving existing system and leads to generation of new methods and techniques for making better violence detection system. The proposed architecture includes mainly two steps: Object tracking and behavior understanding for detecting violence. By using feature extraction process key features (speed, direction, centroid and dimensions) are identified. These features help to track object in video frame. In our approach, we consider two feature vectors namely Violent Flows (ViF) and Local Binary Pattern (LBP) and then Linear SVM is used to classify video as violent or non-violent.