Nigerian License Plate Number Recognition System
Authors: Obayuwana A, Ero OE, Idemudia E
Number plate recognition system usually varies for different countries. Thus, getting a single algorithm/method to accurately detect and recognize all number plates may be rather difficult. This paper presents a methodology to detect, localize, and recognize the presence of a Nigerian vehicle number plate in a picture frame. The procedures used in the image preprocessing stage involved image resizing, converting the RGB image to gray scale image, dilation and erosion, noise reduction, image sharpening, median filtering, edge detection, thresholding and contour analysis. Character segmentation was carried out with the aid of connected component analysis while blob extraction technique was used to extract the characters from the plate using some heuristics. Character recognition was achieved using Artificial neural network with back-propagation learning algorithm. A security access control system was developed using Microsoft Visual C# and the .NET Framework 4.0 to test the algorithm. In the application, the recognized plate numbers were compared with a list of known license plates in a database. If the recognized licensed plate is on the list, access is granted to vehicle otherwise access is denied. The algorithm was able to carry out number plate detection and localization under various environmental and lighting conditions with an average processing time of 0.872 s.
Affiliations: Department of Computer Engineering, University of Benin, Benin City, Nigeria.
Keywords: Image Detection, Image Processing, Character Recognition, Artificial Neural Network, Character Segmentation
Published date: 2019/06/30