Authors: Asad Naeem , Irtiza Hasan , Bilal Haider
Car number plate recognition systems are employed to automate surveillance, access control, vehicle abuse prevention, help law enforcement agencies etc. The development history of license plate recognition systems dates back to 1976 and the first arrest of a stolen car through detection was made in 1981 in UK. These systems work well in the first world countries where car plates are standardized by law. There are many countries without any strict law for plate formats and no generic number plate recognition system exists that solves the localization and number plate recognition for all multi font, multi colored plates. This paper introduces a comprehensive method for the recognition of generic car number plates. Many researchers in the past have tried to address this issue but most of them either focused on specific formats or limited their research to specific color license plates. The Proposed system in this paper adopts a hybrid approach for the localization and recognition of car number plate. The composite approach allows the proposed system to be more adaptive and robust in various conditions without losing the accuracy especially in the localization module. The Recognition of license plates is carried out through different classifiers in order to enhance the overall efficiency and precision of the system. In contrast to other techniques, this system is not only more effective and reliable but also operates accurately and efficiently even under poor illumination and low resolution images.
Keywords: component; Number Plate Recognition and localization, MSER, Shape Analysis, Aspect Ratio, character recognition.