License Plate Recognition: A Comparative Study on Thresholding, OCR and Machine Learning Approaches

摘要

License Plate Recognition (LPR) aims to locate and extract vehicle plate information captures from images or videos. In this paper our objective is to bring forth a comparison based upon the considerations like average accuracy, precision and recall between algorithms according to threshold values, character recognition. The system thus formulated captures real-time input image. It identifies the license plate from extracted image. The work presented in this paper mainly focuses on classification and recognition of characters using Viola Jones Machine learning algorithm. LPR is the most interesting and challenging area of research due to its importance to a wide range of commercial applications, ranging from automated payment services (e.g. Parking and toll roads payment collection), traffic related applications such as road traffic monitoring, searching of stolen vehicles, airport gate monitoring, speed monitoring for more critical applications, to border crossing security and traffic surveillance systems.