LICENSE PLATE DETECTION USING OPECV-MACHINE LEARNING

Abstract

Automatic License Plate Detection and Recognition (ALPR) has become an essential technology in intelligent transportation systems, smart parking management, toll collection, traffic surveillance, and law enforcement. Manual monitoring of vehicle registration numbers is time-consuming and prone to human error, especially in high-traffic environments. This paper presents an enhanced Automatic License Plate Detection and Recognition System using OpenCV and EasyOCR to accurately identify and recognize Indian vehicle registration plates under varying environmental conditions. The proposed system employs image preprocessing techniques such as grayscale conversion, bilateral filtering, adaptive thresholding, and edge detection to improve image quality before license plate localization. Contour-based analysis and geometric filtering are used to detect the plate region efficiently. The localized plate undergoes perspective correction and character extraction before Optical Character Recognition (OCR) is performed using EasyOCR. A rule-based validation module verifies the recognized registration number according to the Indian vehicle registration format, reducing recognition errors. Experimental evaluation on a diverse dataset containing images captured during daytime, nighttime, different weather conditions, and varying camera angles demonstrates high detection and recognition accuracy while maintaining low processing time. The proposed approach offers a cost-effective and computationally efficient solution that can be deployed for real-time traffic monitoring, parking automation, and vehicle access control systems.

Keywords