Work Completion Seminar: Research and Development of a Learning-Based Vehicle License Plate Detection Algorithm

Dear all,

You are most welcome to attend the Work Completion Seminar for Mr Ho Wing Teng (Master of Computer Science), on his research title of Research and Development of a Learning-Based Vehicle License Plate Detection Algorithm. Details are as follow:

Date: 21/03/2011 (Mon)

Time: 11:00am-12:00pm

Venue: 5A meeting room, 5th floor, SA Block, FES, UTAR KL Campus.
Research title: Research and Development of a Learning-Based Vehicle License Plate Detection Algorithm

Candidate: Ho Wing Teng

Abstract:

The objective of this project is to research and develop a learning-based license plate detection framework with Spatial Horizontal Edge Variation features. As the growing of vehicle usage and the importance of automation in license plate recognition, license plate detection is one of the essential modules in the automation system. There are many existing algorithms available in the market in finding the license plate; in this project we introduce a framework that will learn from the image dataset and generate a set of classifiers to detect the targeted object. We tested our proposed framework on Malaysian vehicle license plate detection, and further experiment on Text detection. Spatial Horizontal Edge Variation feature is a simplified edge feature that can extract the edge information of the object through the horizontal orientation to differentiate between license plate and non license plate. Our framework selects a set of best features from a large set of feature set and generates a linear tree classifier to perform the object detection. From our experiments and analysis, we found that vehicle license plates have similarity with text that consist several characters. The changes of the edge information from the left to right horizontally through the license plate can actually provide useful information to distinguish whether it is a license plate or otherwise. In our project, we did further testing on the framework for detecting text from natural scene and side view car detection. For our license plate detector, we tested with some of the Malaysian vehicle license plate and achieve recall rate of 0.93796 and precision rate of 0.55366.

--
regards,
Yong Haur

TAY Yong Haur,
Associate Professor,
Head of Programme (Master of Computer Science)
Department of Internet Engineering and Computer Science (DIECS),
Faculty of Engineering and Science (FES),
Chairperson, Centre for Computing and Intelligent Systems (CCIS),
Universiti Tunku Abdul Rahman (UTAR),
Kuala Lumpur Campus
Jalan Genting Kelang, 53300 Kuala Lumpur
Malaysia.
Tel: (603) 4107 9802
Fax: (603) 4107 9803
Email: tayyh@utar.edu.my or YongHaur.Tay@gmail.com
URL: http://staff.utar.edu.my/tayyh/
URL: http://www.utar.edu.my/