An Implementation Of License Plate Recognition Method Based On Neural Network

With the rapid economic and social developing in the modern life, we need efficient, simple, intelligent management in many areas. For being comfortable, when take a traffic travel trip, we often choose the cars firstly. But the cars driving on the road causes numbers of problems in regulation. The license plate is the accurate way to identify a car, therefore there is a necessary for us to discuss and learn of the technology of license plate identification. Many domestic and foreign scholars in this area have done a lot of experiments and exploration. It produces many algorithms and thought currently corresponding to every step which is made in different processing environment, so the efficiency of each algorithm exist some differences. This article by collecting a large number of Chinese plate images compared a series of algorithms and procedures for the identification of the car license plate and designed a overall flow for the use of vehicle license identification based on the C++ and the Open CV. It can be divided into: image pre-processing of the license, the license for positioning and the character segmentation and identification process in the image. We have done a detailed discussion when introduced every chapter involving the main use of technique and principles. According to the full text of arrangements, they are summarized as follows:Pre-processing step before the image is a key part for the following process, because it filter out so many confounding factors in images which provides a convenient for our next step like identification. In this paper, we use the Gaussian blur technology, binarization, edge detection operator of Sobel and morphological processing to do a detailed introduction. In the positioning process, we use the method based on HSV color space which combined the inherent color characteristics in license models feature and the integrated ones. It could reduce the external interference later providing a help for us. To select the real license plate from so many candidates, we need to carry out before the identification of those who are inclined to judge and angular correction. For many pictures of the identification, we use support vector machine to learn and train them. In the handling of the characters, we introduced firstly the existing segmentation and identification algorithms, and then combined with the implementation of this article in Open CV. It includes the steps of the normalized for the character region、the use of character segmentation of the vertical projection and the character identification based on MLP model of ANN based.In this paper, through a series of steps to learn and realize the identification, I have a further understanding of the purpose. To improve the accuracy and real-time of the identification, the learning will lay a good foundation in the next.