Study Of Image Super-Resolution In Video License Plate Recognition

License plate (LP) recognition is an important part of intelligent transportation system (ITS). When the LPR system is employed to identify fast-moving vehicle, the captured image will be degraded due to the reasons such as blur and the limit of camera resolution. In order to improve the recognition rate, usually, the super-resolution reconstruction (SR) technology is applied to enhance the plate image.This paper reviews the technology of LP enhancement and super-resolution. The main contributions contain the following points. First, analyzing the reason of degradation, and establishing the degradation model, and analyzing the technical difficulty of the SR; Second, as the using the gray projection algorithm to estimate the integer pixel motion vector, and computing the sub-pixel motion vector through Taylor series expansion. Third, considering the drawback of the traditional method, we propose a robust irregular data fusion method; At last, proposing an edge-preserved image restoration.Experiment results show that, the proposed motion estimation method has good performance in precision and computation speed. The method of super-resolution shows robustness to both simulate and actual license plate image.