Localization Method for Camera Networks in Surveillance System

Localization Method for Camera Networks in Surveillance System

The significance of prevention and mitigation of critical issues especially in the homeland security has been increasing day by day. Emergence of autonomous video analytics tools greatly helped in the prevention of security threats. The recognition of video analytics for anomaly detection based on a set of unsupervised approaches has many fundamental technical challenges. This entails autonomous object localization and tracking technique especially in the presence of occlusion. This paper focuses on deriving a solution for the object detection and tracking in a heterogeneous camera network. The object tracking method is mainly based on Kalman filter whereas frame difference algorithm is used for object localization. This detection and tracking solution is expected to significantly reduce the effect of occlusion while tracking the anomaly. The organisation of the thesis is done into various chapters. The first chapter contains an introduction to the video surveillance system and the need for an unsupervised approach. This chapter also states the objective of the research. The solution overview gives high level solution architecture of the proposed system. The second chapter focus on the literature overview in which the citation from different papers in the field of video analytics, Kalman filter implementation and camera configuration has been referred. Chapter 3 provides the methodology in which a brief introduction to the basic algorithms used in the solution, the Kalman filter and the frame difference algorithm, are discussed. This is followed by the solution architecture of the proposed system. Chapter 4 shows the Matlab implementation of the mentioned algorithms. In Chapter 5, the results of the implementation are discussed. Chapter 6 talks about the summary of the work done and conclusion. This chapter also includes the future enhancements suggested.

Jose Chalissery Benita

Computer Science