Design And Realization Of The Vehicle License Plate Localization

The vehicle license plate recognition is one of the important components of the modern intelligent traffic system, It is one widely application everywhere. With the rapid development of society, it is must be considered that the vehicle license plate recognition was, It plays an important role in in the modern intelligent transportation system.This thesis mainly studies the vehicle license plate localization algorithm design and implementation of vehicle license plate location, and the vehicle license plate localization is the key and important part of the vehicle license plate recognition system.The vehicle license plate recognition system mainly includes the following three parts:vehicle license plate pretreatment and location, character segmentation and character recognition. Paper were discussed about research on vehicle license plate recognition and information processing of each process technology, and were discussed in the vehicle license plate pre-treatment, location and recognition process in treatment technology, focusing on the vehicle license plate Localization arithmetic design and implementation of Matlab.This thesis mainly works and contributions include as the following several aspects:First, Design and implement the license plate preprocessing algorithm based on image enhaucement and morphological image processing. In the first place, it will be deal with gray processing of this algorithm, and to be filter out the noise by the scale filter, and smooth the image in order to enhance the image effect by the fixed threshold method, then evaluate by the brightness, at the same time using the method of mathematical morphology to treat different situations, finally, after the smooth image processing for larger particles problem again.Second, Designed and implemented license plate localization algorithm based on prior knowledge and edge detection.According to China’s license plates characteristics of length-width ratio and physical area, the vehicle license image will be coarsed locating by prior knowledge of this algorithm, then preprocessing of the vehicle license plate image by area with taller than initial positioning, and then the images using brightness detector for positioning detection, finally,by the image distance ratio distinguish the vehicle license plate area.Finally this thesis adopted roberts operator to be detected by double edge of image. According to the size of the real license plate images, it include three conditions for processing. Third, Design and implement the license plate positioning algorithm based on the projection method.The algorithm considering the filming and tilt problem, it will be processe vehicle license plate area according to the structure and texture characteristics. At the same time avoiding the unconnected and adhesion background environment fault, reasonable selection of the shape and size with the horizontal and vertical structure, excluding not conform to the requirements of the connected area, improved the accuracy of localization. So, it need to filtered candidate area after the coarse vehicle license plate locating,and the deformation correction processing later, frame and rivet removal, and accurately localization to the license plate’s character area high and the the border of the characters left and right side.Finally, the Matlab simulation platform, successful implement the simulation experiment of vehicle license plate accurate locating positioning. And accurately locating the vehicle license plate and interception of license plate area and then display output result.