Super-resolution Reconstruction Based On The Image Of Vehicle License Plate

As a result of motion blur, point spread blur, under-sampling and sensor noise and so no, the image will occur degraded during acquiring. In order to improve the quality and spatial resolution of an image, super-resolution reconstruction techniques are commonly used.This paper reviewed the relevant theories and classic algorithms about super-resolution reconstruction, used Bayesian and projections onto convex sets as the main mathematical tool, and discussed in detail the application of super-resolution reconstruction of vehicle license plate image. The main work of this paper was listed as follows: established the degraded model on the basis of discussing in detail a variety of degraded factors existing in the image acquiring, analyzed the feasibility of super-resolution reconstruction; discussed the single-frame image reconstruction based on interpolated methods, and compared the running time and peak signal to noise ratio of these methods; analyzed the performance of several motion estimation algorithms based on block in the search speed and the quality of motion-compensated image; analyzed the basic principle and implementation steps of IBP, POCS, MAP, analyzed and compared the advantages and disadvantages of various reconstruction algorithms in the special application of license plate image reconstruction by experiment.The results of experiments indicated that: for the computer simulation and traffic monitoring video images, reconstruction algorithms based on multi-frame can reconstruct a high-resolution image of higher quality, and provide more information than the single-frame reconstruction algorithms.