Automated License Plate Recognition

Abstract

A vehicle can be identified with its license number. This number is generally used for many applications. So a method to find out the license number automatically is very much necessary. License plates were usually extracted from static images of vehicles by clustering and then the characters are segmented using dilation process. But this static image method does not give very accurate results when the license plates are of different sizes or different color. It could not succeed in recognition of license plates where the characters are broken and illumination changes. So, the proposed system is designed in such a way, to solve these problems. It takes a real time image as input. License plate can be detected from the input by detecting edges using Sobel edge detection technique, followed by morphological operations. The detected license plate can be extracted and the characters in the license plate are segmented by connected component analysis. This is a faster and more accurate approach. The segmented characters can then be matched with character templates stored in the database and the correct character is recognized. Images of vehicles, taken at real time at various angles having different backgrounds and illumination conditions license plates of any color or size and characters of any font style and language are recognized.