Image super-resolution via sparse coding for Chinese license plate recognition

Some plate recognition systems can not recognize the low-resolution license plate images correctly because the resolution of the input plate image is very low. The proposed method in this paper enhances the resolution of the Chinese license plate image via sparse coding. It employs dictionary learning to get the optimal overcomplete dictionary pairs and introduces the regularization terms to recover the high-resolution plate image. Experiments show that the plate image recovered by the proposed method can be well recognized. That is, the proposed single image super-resolution method in this paper can promote the plate recognition effectively.