Skew correction process in automatic number plate recognition (ANPR) system

Abstract

In some situation the vehicle license plates are detected with different inclinations, in such situations it is difficult to obtain the accurate alphanumeric characters available on license plate even after proper recognition. Hence, in such situation a system is required which can detect an image with high degree of proper orientation and it needs to be corrected before proceeding for recognition. “Automatic Number Plate Detection and Recognition” (ANPR) is a real-time system, which is mainly used in countries where degree of orientation is greatly observed and number plate standards are hardly followed. In order to overcome such situation of skewness, one can follow skew correction methodology. In ANPR methodology, skew correction is one of the modules and it is achieved in three steps. The first stage contains three parts i.e., masking, thresholding and finally the contour detection. The methodology for corner detection of extracted license plates is described in second stage which uses the Douglas-Peucker algorithm and in third stage perspective transformation and Homography calculations is used to achieve the correct inclination of license plate. The main aim of the proposed method is to build a system with simple methodology to achieve proper inclination of the license plate using corner detection related algorithms and Homography matrix computation.