Actual License Plate Images Clarity Classification via Sparse Representation

摘要

The quality of the license plate image has a great influence on the license plate recognition algorithm. Predicting the clarity of license plate image in advance will help the license plate recognition algorithm set appropriate parameters to improve the accuracy of the recognition. In this paper, we propose a classification algorithm based on sparse representation and reconstruction error to divide license plate images into two categories: high-clarity and low-clarity. We produced over complete dictionaries of both two categories, and extract the reconstruction error of the license plate image that to be classified through the two dictionaries as the feature vector. Finally we send the feature vector to SVM classifier. Our Algorithm is tested by the license plate image database, reaching over 90% accuracy.