A Robust Segmentation Free License Plate Recognition Method

Abstract

Automatic License Plate Recognition (ALPR) aims to extract vehicle license plate information from an image/video. Different factors associated with license plates, such as low-resolution, non-uniform illuminations, and view angle add a level of difficulty for any ALPR algorithm to perform accurately under aforementioned situations. To address these issues, this paper presents a robust and efficient segmentation free ALPR algorithm. The proposed algorithm uses adaptive boosting integrated with the Linear Discriminant Analysis (LDA) for features extraction and Classic Nearest Neighbor Classifier (CNNC) for classification. Detailed simulations on the Caltech database reveal that the proposed method outperforms recent state-of-the-methods in terms recognition accuracy, precison, and recall ratios at different day times on different angular views license plates.