De-blurring methodology of license plate using sparse representation

Abstract

A super-resolution reconstruction from single image algorithm designed for license plate recognition is proposed in this paper. Low resolution images database is generated by down sampling and adding white Gaussian noise to the super resolution license plate database. The low-resolution image can be viewed as a down sampled version of a high-resolution image, where its patches are assumed to have a sparse representation with respect to an over-complete dictionary of prototype signal atoms. The principle of compressed sensing ensures that under mild conditions, the sparse representation can be correctly recovered from the down sampled signal. Therefore, two dictionary of low and high resolution from same images patches are trained. Finally, super resolution images from single low resolution image are recovered, by solving an optimization problem by genetic algorithm.