A Centroid based Hough Transformation for Indian license plate skew detection and correction of IR and color images

Abstract

Skew Correction is a processing-stage between LP Localization and Character Segmentation in License Plate Recognition system used to identify a vehicle by its number plate. License plate is skewed in captured image due to the positioning of the vehicle with respect to the camera while capturing the LP image. The skewed license plate affects badly on the accurate character segmentation and recognition. After localization skew correction technique is applied in order to get correct character segmentation followed by character recognition. In this paper, a Centroid based Hough Transform technique is presented for skew correction of license plate which performs better than the other approaches of skew correction including Thresholding, Connected Component analysis, Hough Transform and Centroid method. The performance of the proposed algorithm has been tested on live captured LP images, yielding better performance in car license plate segmentation and hence the presented algorithm is good for all types of LP recognition applications due to its direct and simple approach with minimal computational time.