Universiti Teknologi Malaysia Institutional Repository

A computer vision system for the classification of moving object

Mohammed Osman Saleh Bilal, Sara (2005) A computer vision system for the classification of moving object. Masters thesis, Universiti Teknologi Malaysia, Faculty of Electrical Engineering.



The aim of this research is to produce a system that can detect the moving object and classify it into three classes: “Humans, Vehicle and Animals”. Using fixed video camera in outdoors environment, the system will capture the images and digitize them using (Piccolo Pro II) frame grabber at a rate of 25 frames per second. The Background Subtraction technique has been employed in the work as it is able to provide the most complete feature for data. However, it is extremely sensitive to dynamic changes like changing of illumination. Background Subtraction is done by taking the differenc e between any frame and the background in detecting the Moving Object. In order to reduce the effect of noise pixels resulting from the Background Subtraction operation, a number of pre-processing methods have been applied on the detected moving object. These preprocessing operations involve the use of median filter as well as morphological filters. Then the outline of the object will be extracted using border extraction technique. The classification makes use of both the shape and the dynamic features of the objects. In increasing the performance of the classification, all features are sequentially arranged so that the goal of this research is to be achieved. In this work, the performance achieved is 93% for class human, 93% for class vehicle and 64% for class animal.

Item Type:Thesis (Masters)
Additional Information:Thesis (Master of Engineering (Electrical-Electronics and Telecommunication)) - Universiti Teknologi Malaysia, 2005 ; Supervisor : Assoc. Prof.Dr. Sued Abd Rahman Al-Attas
Subjects:T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:Electrical Engineering
ID Code:3000
Deposited By: Fazli Masari
Deposited On:28 May 2007 03:10
Last Modified:25 Jun 2018 00:45

Repository Staff Only: item control page