A system for interactive surveillance video retrieval

• Main Author: Yang, Wujun
• Other Authors: Chan Kap Luk
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: http://hdl.handle.net/10356/64809
• Institution: Nanyang Technological University

An interactive Surveillance Video Retrieval System is a software system that manages retrieval of surveillance videos from a structured database. The system can be extended and applied widely in many areas such as Crime Investigation, Immigration Checkpoints, Subways and Traffic Control. The recorded videos from surveillance cameras are first processed through various image processing techniques. The process aims to understand the content in a scene and to extract useful information for further processing. Some common content of our interest is the individual object that is in the scene and other information associated with it such as the size, color and movement. The information are stored in a database and the file system for future retrieval. The retrieval software provides various ways to obtain the list of desired videos. In short, it gives a search functionality through filters. This dissertation reports the work done to design and develop such a system with functionalities similar to what just been described. In this system, the targeted sets of surveillance videos are provided, which can be the video recorded from any surveillance cameras in real life. A backend processing program detects moving object and records the movement and other information of each moving object. The information captured is stored in a database and retrieved through a combination of filters from the retrieval software system. The replay of the video is organized in a few categories where user is able to access the desired part of the video conveniently. In the testing of the surveillance video retrieval system, it is found that the video processing methods affects the desired performance of the system. In this dissertation, a study was conducted to understand how the different video processing methods worked in videos with different scenes and different sizes of surveillance targets. The dissertation includes an introduction of the principles of the selected video processing methods and comments on the performance of the methods for the different types of videos.