Based On Mathematical Morphology License Plate Location And Recognition Technology

With the development of modern transport technology, automobile has become one of the indispensable means of transportation in the daily life, which means more requirements in traffic management. With the development of the automation of the charges, parking management, etc on the highway. License Plate Recognition (LPR) technology has become a hot topic.LPR system is an intelligent traffic management system based on image processing and character recognition technology, which can intake the vehicle license plate recognition and image automatically. A complete LPR system is made up of different units such as the license plate location, characters division and character recognition.The project puts forward a set of simple and efficient method, based on mathematical morphology of the license plate location and recognition. Location comprehensively considered the gray and the edge information, as well. First of all on the pre-image using the gray-scale transformation, noise and enhancing contrast, edge detection, binary, and other methods of image processing, improved image quality, strengthen the image of the region; Then, use a binary image based on the characteristics of gray-scale changes in the positioning method, according to the gray-hopping search license plate region, making use of the license plate geometry characteristics of the candidate screening area, we can get the location of the license plate. It has been proved by the experiments that the algorithm is simple, efficient, practical, fast positioning, and it's effective for the acquisition of the image database, which rarely has occurrence in the basic positioning error, or cuts off the edge of the effective information. Characters division improves the traditional projection method, using the binary images of white pixels vertical projection method, in the projection in the horizontal calculation of the largest interval position, that is, the location is the first character, the position between the character 2,3; based on this location, more or less in accordance with the width of characters of the left divided by 2 characters, a division of the right 5 characters. The advantage of using this method is making full use of the features of the plate layout, making it well separated from each of the characters. Recognition is based on the use of a template matching method, combined with statistics, that is, using the existing features of the vector, to be identifiable from the image to extract images of a number of features and characteristics of the template corresponding to the amount of comparison, the calculation template Vector and the similarity between the current vector, and then present it with the characteristics of the vector into the highest similarity template for a class. We can use the minimum distance method to calculate the template image and the distance between features, and pattern recognition by the k-neighbor method, the method can effectively complete the resolution of different fuzzy and the license plate to identify the extent of the work.Experiments show that this set of approach has a higher degree of recognition, which can lead a satisfied effect.