A Proposed License Plate Classification Method Based on Convolutional Neural Network

Nowadays convolutional neural network(CNN) has been successfully applied in image processing. At the same time license plate recognition is more and more universal. Before recognition with deep learning, we need to collect enough images to train the network. The quality of data is very important. The current methods of getting license plate based on deep learning are increasingly various, however, there are still many images where illumination, size and blurriness make it is extremely difficult to recognize. As a result, images with low quality eventually affect the accuracy of recognition. Therefore, license plate classification is essential to eliminate low quality images so that improve the quality of the dataset. In this paper, a method based on CNN is proposed to deal with license plate classification. We use a seven layers CNN and ultimately the best result reached 98.79%.