The recognition of the closed-set patterns in the Saudi license plates

In this paper, four closed-set patterns in the Saudi license plates can be recognized using our system. The four closed-set patterns are the numbers and letters in both Arabic and English languages. The system is based on classifying each closed-set in a single decision tree. We applied different feature extraction methods in order to analyze each possible line crossing the images. We studied the performance of the classification by adding the Gaussian noise to the license plate images, and finally we proposed a method in order to enhance the recognition accuracy with or without the image noise.