Application Of Human Detection Algorithms In Fixed Video Surveillance System For Reserved Places

Abstract

Insecurity in many parts of the world is a major challenge that has led to increased deployment of video surveillance in both private and public entities. As a result, the need to have automated video surveillance systems is receiving serious attention as demonstrated in intense ongoing research in Human Detection Algorithms by various research communities in the field of computer vision. In this dissertation we propose an automated fixed video surveillance system based on human detection algorithms with a focus on wildlife surveillance. Today, endangered species of wildlife, in particular the Rhinos and Elephants are faced with the threat of extinction due to poaching, which is motivated by the value of their horns and there is need to use video surveillance technology to supplement current methods of wildlife protection. We used prototyping method to implement the system using Open CV library algorithms in MATLAB language by cascading upper body, full body and face detection algorithms to achieve the objectives set out in this research. The proposed system successfully detects human beings in different postures such as upright facing the camera and away from the camera, walking upright, bending, sitting and squatting. With average positive detection rate of 90% achieved in different test environments, it has been demonstrated that the system can be used to improve the efficiency of video surveillance of reserved places.