Automatic vehicle license plate recognition with color component texture detection and template matching

Abstract

Automatic vehicle license plate recognition (LPR) is important for intelligent traffic surveillance systems. This paper suggests a vehicle license plate algorithm, color component texture detection and template matching (CCTD-TM). CCTD-TM has advantages of ease of implementation and highly efficient in calculation. We suggest a novel algorithm of color component texture for license plate localization. This algorithm takes advantage of the feature of fixed color texture of plate base and character. The image preprocessing and character recognition by template matching parts are included in the LPR algorithm. The preliminary results demonstrate an average detection rate over 96.5% and an average recognition rate over 89.9% on hundreds of vehicle images tested in the experiments.