Automatic Number Plate Recognition System

Abstract

During the past few years, as number of vehicles has increased on road, managing traffic has been a big challenge for government. To catch and punish traffic offender, lots of manpower is required. The automatic surveillance system can make this job easier. As number plate is only mean to identify vehicle, Automatic Number Plate Recognition (ANPR) is the substantial part to make system automated. ANPR is a system in which every vehicle passing through the view of camera, are monitored. As this project is for Indian vehicles, the challenge in this project is to recognize number plate because no standard for number plate size or number font type and size are followed. The thesis is divided into three modules: Finding the optimal frame with better probability of finding the number plate from video, Number plate detection and Character Recognition. For the first module the maximum-difference method is used. This method works when there is a single vehicle in the view at a time. The second module, Number plate detection, is implemented based on the property that there are more vertical edges near number plate area. So the edge-based method, which finds horizontal strip containing number plate from the image, is used for number plate detection in conjunction with the blob-based method, which finds and segments the number plate characters. This method gives promising results. The segmented characters are forwarded to the third module, character recognition. After the recognition of number plate characters the numbers are stored for further use.