Automatic number plate extraction based on wavelet transform and bounding box technique

Abstract

Automatic Number Plate Recognition (ANPR) technique occupies a significant importance as an intelligent management and monitoring system in real time. It acts as an essential part in a large portion of the movement related applications, for example, activity administration, auto stopping/outstay control, tracking and identification of undesirable or un-approved vehicle in the region of profoundly secure structures. An extensive variety of research on ANPR frameworks are accessible in writing. But these existing algorithms become inadequate due to changing illumination level and background conditions. Non-standard design of number plates in Pakistan make it even harder for the existing systems to show promising results. This paper concentrates on a productive framework to design an efficient ANPR system by applying Two Dimensional Discrete Wavelet Transformation. Haar wavelet is being proposed due to its simplicity and being computationally convenient among all wavelet. Wavelet transformation along with bounding box method promise a system with better performance in edge detection and character recognition.