Automatic number plate recognition system by character position method

Automatic number plate recognition (ANPR) is an important image processing technology used to recognise number plates of vehicles. Plate localisation and character recognition are two stages of ANPR. In this paper, a methodology has been proposed to develop robust ANPR system. A new algorithm has been proposed for number plate localisation which is based on character positioning method. Character recognition is done with support vector machine in which feature vector is calculated by recursive sub-divisions of character image. The problem of similar shape characters has been solved by syntactic analysis of number plate format for a particular geographical region. The system has been tested on 419 sample images from various countries with various variations in viewing angles, illuminations and distances. Experimental results show that the proposed system detects number plates and recognise characters successfully. The overall success rate of plate localisation is 97.21% and recognition of number is 95.06%.