Automatic Number Plate Detection using Haar-Cascade Algorithm Proposed for Srilankan Context

The Automatic Number Plate Recognition (ANPR) is a core concept in Intelligent transportation management systems. This paper presents a methodology relevant to Srilankan context for detecting the number plate from real time video and character recognition from the number plate image identified. We proposed, pre-trained Haar-like Cascade Classifiers to detect the number plate region by analyzing the video, frame by frame and locate the number plate using a bounding box. The detection module also involves an Ensemble methodology of boosting supported by AdaBoost algorithm to optimize the detection process. The Character recognition from the number plate has been achieved through Tesseract OCR after preprocessing the number plate image. This model works with an outstanding accuracy level in detecting the number plates as well as in recognizing the characters in the number plate. The model also provides us satisfiable result in exceptional situations like adverse lighting and weather conditions and changes in camera angles.