D-PNR: Deep License Plate Number Recognition

Abstract

Deep learning empowered the license plate recognition (LPR) system which can efficiently extract the information from a vehicle’s license plate. LPR has various applications in this digital world. Due to technology growing at a rocketing pace, there is a rapid growth in the number of vehicles on road. Even self-driving cars are soon to be a common sight. This is causing a fast and frequent growth in the accidents occurrence and other mishaps. Thus, there is a need for monitoring traffic and security surveillance. LPR technique is not new. However, traditionally the extracting features from an image/ license plate were done hand-tunned which make the recognition process time-consuming and error-prone. In this paper, we proposed a novel machine learning approach to recognizing the license plate number. We used one of the most successful deep learning method, convolutional neural network (CNN) for extracting the visual features automatically. Suitable localization and segmentation techniques are employed before CNN model to enhance the accuracy of the proposed model. In addition to this, the D-PNR model also takes care of proper identification from images those are hazy and is not suitable-inclined or noisy images. Qualitative and quantitative evaluation is done in order to compare the performances of the proposed D-PNR model and state-of-the-art models. A computing analysis of our approach also shows that it meets the requirement of the real-time applications, i.e., monitoring traffic and security surveillance