Brazilian Mercosur License Plate Detection and Recognition Using Haar Cascade and Tesseract OCR on Synthetic Imagery

Abstract

AbstractThe increase in the number of vehicles and consecutively in traffic on roads and highways bring the need to create systems for the detection and recognition of license plates. The new Mercosur license plate model adopted in Brazil is still in transition, so a low number of vehicles with the new license plate are in circulation. This article proposes to present Brazilian Mercosur license plate detection techniques on synthetic imagery and character recognition using Cascade Classifiers and Tesseract OCR, respectively. For license plate detection, a multi-stage classifier is trained with Haar-type features using the Cascade Trainer GUI tool (Version 3.3.1). The proposed method was validated using the Mercosur license plate dataset on synthetic imagery. It was obtained 83.82% of detection success and 95.72% digits classification accuracy rate.KeywordsLicense plate detectionDigits recognitionVehicle license plateHaar cascadeTesseract OCRImage processing