Research On Super-resolution Reconstruction Algorithm On The Image Of Vehicle License Plate

With the improvement of living standards, the popularity of the vehicle led to accidents frequently occur, but due to the low level of imaging hardware technology used to shoot, resulting in sometimes can’t get a clear picture of the license plate. This paper from the software aspects of the license plate image process,for improve the license plate image resolution by super-resolution image reconstruction techniques.In this paper, based on observation model for the license plate image of the super-resolution reconstruction. First, image registration, a detailed analysis on the basis of Keren Registration Method combination of Gaussian pyramid improved method; secondly a detailed description of two common spatial algorithms, namely POCS algorithm and MAP algorithm to achieve a license plate image super-resolution reconstruction, and the advantages and disadvantages of the two algorithms for detailed comparative analysis; and finally, because of the super-resolution image reconstruction is an ill-posed problem, regularization theory by introducing a super-resolution image reconstruction into a solvable the well-posed.In the above regularization theory, based on the total variation model based on the shortcomings of the proposed adaptive total variation based image super-resolution reconstruction of the improved method, the first use of a median filter for image pre-processing, overcoming total variation ladder effect model was introduced; re-use image information, adaptive selection of the parameters in the diffusion equation, so that the image to remove the salt and pepper noise while maintaining a clear edge, as far as possible in order to make the resulting image faithful to the original image, the last to join loyalty items. Simulation results show that the effect of the proposed algorithm is obvious that the human eye can feel the direct effect of improving.