Vehicle License Plate Recognition Using Pixel Information

Bedrossian, Sandra (2019) Vehicle License Plate Recognition Using Pixel Information. Bachelor's Thesis, Artificial Intelligence.

Preview	Text AI_BA_2019_SANDRABEDROSSIAN.pdf Download (2MB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (143kB)

Abstract

Law enforcement and public safety relating to the recognition of vehicles from image or video footage call for the efficient and rapid performance of Automatic License Plate Recognition (ALPR) systems. ALPR systems are devised by an integration of vehicle localisation, license plate detection, license plate segmentation and character recognition. This project primarily aims to extend the ALPR system devised by Kasaei et al. (2011) to discover character recognition algorithms that yield a reliable accuracy in the recognition of Persian license plates. Specifically, the performance of template matching, Gaussian-weighted template matching, naive Bayes and a multilayer perceptron within the domain of character recognition are compared. Preliminary results while testing on a data-set of plain characters via 10-fold cross validation indicate the multilayer perceptron outperforms naive Bayes, Gaussian-weighted template matching and template matching, respectively. Further testing on a dataset of 500 highway images of vehicles indicate contradictory results with template matching marginally outperforming the multilayer perceptron, naive Bayes and Gaussian-weighted template matching respectively. The differences in accuracies are however generally close with large standard deviations and should not be exaggerated; further preprocessing of the segmented characters prior to recognition should steer results to those more representative of the preliminary results on the characters dataset.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Mohades Kasaei, S.H.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	25 Jul 2019
Last Modified:	26 Jul 2019 06:36
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/20437

Actions (login required)

View Item