Moving object detection and classification for smart surveillance system /

Public surveillance monitoring has attracted a lot of attention in the last few years especially in Intelligent Surveillance System. Street crime is increasing in recent years, which has demanded more reliable and intelligent public surveillance system. The surveillance systems such as CCTVs are che...

Full description

Saved in:
Bibliographic Details
Main Author: Mohammad Ariff bin Rashidan
Format: Thesis
Language:English
Published: Gombak, Selangor : Kulliyyah of Engineering, International Islamic University Malaysia, 2016
Subjects:
Online Access:Click here to view 1st 24 pages of the thesis. Members can view fulltext at the specified PCs in the library.
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Public surveillance monitoring has attracted a lot of attention in the last few years especially in Intelligent Surveillance System. Street crime is increasing in recent years, which has demanded more reliable and intelligent public surveillance system. The surveillance systems such as CCTVs are cheaper nowadays making it easily deployable. However, the dependency upon human to monitor the video feed is very costly and inefficient. Human tends to become bored due to the dull nature of the monitoring activity. Thus, an intelligent surveillance system which could detect and classify moving objects is proposed and developed in this research work. In this study, the foreground segmentation based on adaptive background subtraction is used to extract foreground objects from the image. In the process of classification, the Adaptive Neuro-Fuzzy Inference System (ANFIS) is adopted in order to incorporate fuzzy logic into neural networks framework. This enables the system to deal with cognitive uncertainties in a manner more like humans. In this research, the emphasis is given to the three major processes which are object detection, discriminative feature extraction, and classification of the target into pre-defined classes, pedestrian, motorcyclist, and car. The adaptive network based on Neuro-Fuzzy was independently developed for three output parameters, in which each of the fuzzy inference system constitutes of three inputs and 27 Sugeno-rules. The system was developed in MATLAB simulation environment and performance analysis is done on five real street scene datasets which have different background complexity. Results obtained shows that the proposed system is able to detect the desired moving objects with an average accuracy of 93.67% and able classify for each class with an accuracy of 93.63%. Moreover, the overall accuracy of our proposed surveillance system in classifying street scene objects gives promising results compared to several recent benchmark studies that tested with the well-known dataset, PETS 2001.
Physical Description:xx, 163 leaves : ill. ; 30cm.
Bibliography:Includes bibliographical references (leaves 149-158).