Research On Location And Recognition Algorithms Of Tilt Vehicle Licence Plate

Traffic congestion becomes a common phenomenon in large and mediumsized cities. Now the fixed traffic intersection camera is limited by many factors,so the license plate recognition technology has played a vital role in the vehiclemanagement, which is also the part of China’s "Safe City". This paper researchesthe existed license plate recognition technology, analyzes the type andcharacteristics of the tilt license plate, make a more in-depth research on theoblique license plate localization and recognition algorithm. The goal is toachieve the license plate recognition automation, improve the currentidentification method to increase the inclined license plate tilt angle duringshooting, therefore the existing camera angle range of horizontal and verticaladaptation can improve character recognition rate and speed.First, this paper analyzes the recent researches about the tilt license platelocation and recognition, proposes a tilt license plate location recognition fusionmethod. We obtain a binarized image by pretreatment of the original image,strengthen the image information through the edge detection operator; taking theetching operation on the edge of the binarized image by pretreatment of theoriginal image, strengthen the image information through the edge detectionoperator; taking the etching operation on the edge of the binarized image withmorphology operation; then get the part of the license plate candidate region. Bythe rate of the width to length of the license plate, we can remove interference onthe license plate, finally we find the target license plate and complete tilt licenseplate location. This method can reduce the inclination conditions increases th eeffective angle of the camera monitoring during shooting, increase the shootingof the identifiable range. Then, the acquired target license plate by Radon transform, the rotation correction on the extracted license plate. But thecorrection characters cannot be completely restored. For getting the accurateidentification of the characters, this paper analyze the advantages anddisadvantages of various recognition algorithms, at last using the templatematching method in the original template on the basis of its rotation and scaling.This approach can not only get tilt character recognition results, but also have aprominent effect, to some extent improve the identification efficiency andaccuracy of identification.Finally, this paper based on Matlab to be involved in all steps of the licenseplate image preprocessing, license plate location, license plate recognitionalgorithm simulation results and analysis are given in order to verify thecorrectness and validity of the proposed algorithm. The experiment shows thatthe fusion algorithm has the positioning accuracy recognition speed and, expandthe existing monitoring camera angle.