Automatic Farsi License Plate Recognition Using Morphological Functions and Artificial Neural Network for Online Applications

In this Paper, a new algorithm for license plate recognition is proposed, which is followed by a neural network in order to recognize the characters. In image processing section the morphological functions are used after edge detection in such a way that the license plate is recognized in a simple way, hence the processing time is reduced. After that, the characters are separated. Each character is entered a three layer feed-forward neural network in order to be recognized. This algorithm is performed over 300 different images, which are different in some point of views such as distance of photography from plate, angle of photography and intensity of luminosity. The algorithm is performed using MATLAB and the results show 93.7% success in license plate recognition and 98.07% success in character recognition by neural network. The results indicate that the proposed algorithm is suitable to work as online license plate recognition system.

کلیدواژه ها:

License Plate Recognition ، Farsi License Plate ، Morphological Operations ، Neural Network ، Edge Detection