Parallel License Plate Recognition Scheme Based On Image Splicing For Intelligent Transportation System

Intelligent transportation system(ITS)is one of the most advanced applications in smart city,which provide a better transportation service with safer,more coordinated and accurate traffic control.ITS integrated various technologies applied,from modern information system such as computer science and communications system to some digital control techniques like sensors technologies in order to achieve a high-efficiency transportation management system.Moreover,as an important role in ITS,Vehicle License Plate Recognition(VLPR)has been widely applied to the modern vision community like traffic surveillance.As so far,many recognition algorithms have been proposed in recent years and have applied to smart parking lot area.Those algorithms are based on four links from image noise reduction and compression;license plate localization and segmentation;character segmentation and recognition to image sink.However,majority of those algorithms have obvious limitations,which are not capable of handling complex environment such as dynamic and cluttered background.For example,if there is numerous license plates need to be processed,the algorithm is no longer a real-time technique.According to it,this paper will present a new method called parallel license plate algorithm that is based on image mosaic.This method could possibly be divided into four steps,which are as follows:1:Target candidate localization:Based on edge detection,localize each target candidate and extract the location’s features.2:Vehicle License Plate splicing:Given w is the image length,h is the image height,and suppose we have N plate image,by applying the image mosaic algorithm,an image with length of w*N,height of h can be obtained.3:Character segmentation and recognition:The projection method will be utilized for character segmentation and SVM classification will be used for training and recognition.4:Data output:7 characters will be output each time,so totally for N times,there will be 7*N characters.Additionally,this report will simulate and analyze the algorithm to confirm if the method is application.