Canny Edges Detection Combined Filtering Algorithms Based License Plate Recognition

The concept of digital image processing is one of the main sections in computer science.Image processing has become a need used in most advanced areas of our society,such as medicine,security,technology,entertainment,and media.For the security matter,especially about traffic and road safety,it is very important to keep tracks of vehicles and be able to identify them in case of road traffic accidents,speeding,violation of traffic rules,etc.by identifying precisely the vehicle license plate number which is an effective way to identify vehicles.This process requires the use of various license plates digital images.Therefore,algorithms for object detection and recognition,their analysis,and computer vision have been developed.Some of these algorithms are based on edge detection.The various edge detection algorithms such as Sobel,laplacian are failed to meet the low area and reduced delay.As a traditional edge detection algorithm,Canny operator is widely used and improved.Canny adopts hysteresis threshold,the different values of threshold have great influence on the detection result,but the number cannot reflect the detection result intuitively.Canny Edge Detection Algorithm gives simple edge detection operation which reduces the time and memory consumption.It is the special algorithm to carry out the edge detection of an image.In this thesis,our proposed approach uses canny edge detection algorithms for the license plate localization in the image combined with some filtering methods for license plate identification and characters recognition and extraction are performed through an Optical Character Recognition system.The Canny edge detection algorithm aims to identify the ROIs find finding contours by performing some image processing operations of image blurring,thresholding by Otsu's algorithm,noise reduction.The boundary of characters in the vertical and horizontal lines was scanned after plate image was converted into the binary system.This method is required to be able to perform in many different conditions and images qualities.This study presents an implementation of a license plate recognition system in order to detect specific areas from images,describes the various steps required to extract text from any image file(jpeg / png),and creates a separate text file that consists of extracted information from an image file.It examines the shortcomings of various image processing applications available and works on overcoming them by performing variable level of image processing and filtering.The OpenCV library,which uses Python language,is used for image processing and Tessaract-OCR is used for characters recognition.A variable level of image processing ensures that different images receive different levels of processing for optimum text results.Test results show that the proposed approach successfully identifies and extracts characters from most test images.A recall rate of 86% was obtained through our method for license plate detection with 87% of accuracy.Nearly 97% of success rate was obtained for license plates characters recognition.