Simple and efficient fuzzy text picture sharpening processing method

Abstract

The invention provides a simple and efficient fuzzy text picture sharpening processing method. The fuzzy text picture sharpening processing method based on the sparsity priori knowledge is provided byanalyzing the specific sparsity priori knowledge of a text picture, restraining the sharpening brightness and gradient of the fuzzy text picture by using an A0 norm, restraining a fuzzy kernel by using an A2 norm and aiming at the global motion blur of the text picture. Before sharpening processing, denoising the blurred text picture, and then solving the A0 priori knowledge sparsification modelby adopting an alternate iteration method to obtain a blurred kernel and an intermediate clear text picture; compared with other methods in the prior art, the method provided by the invention has a better blurred text image restoration effect and is particularly suitable for informatization processing of text images, and after-vibration artifacts are removed through a simple and efficient bilateral residual filtering method.

Description

technical field [0001] The invention relates to a method for clearing and processing fuzzy text pictures, in particular to a simple and efficient method for clearing and processing fuzzy text pictures, and belongs to the technical field of fuzzy text and picture processing. Background technique [0002] With the rapid development of visual information technology and the rapid popularization of digital cameras and smart phones, digital images have become an important form for people to obtain information. The quality of digital images is very important for people to obtain information. A text image is an image containing text or characters. As a special type of image, a text image has a wide range of application scenarios. However, in real life, due to various reasons, many text images obtained by people are blurred. In many cases, people need to recover clear text or characters from blurred text images and extract important information therein. For example, people tend to t...

AI Technical Summary

Problems solved by technology

[0013] One is the mainstream image sharpening method based on the heavy-tailed distribution of ordinary clear images, which is designed based on the gradient of ordinary clear images. It has a good effect on the clearing of ordinary images, but the effect is not ideal for blurred text images. The existing technology cannot be applied to the sharpening processing of blurred text pictures. Due to the great difference between clear text pictures and fuzzy text pictures, the prior art does not have a method for clearing blurred text pictures, and the recovery ability of text Poor, almost useless, existing methods have limitations

[0014] The second is that the existing technology does not use a filter algorithm specially suitable for blurred text image denoising to remove the noise in the blurred text image for the Gaussian noise and impulse noise in the image, and cannot adaptively determine the appropriate size of the rectangular window according to the amount of impulse noise , cannot remove the noise while protecting the edge details of the text image

Unable to self-evaluate the weighted mean filter, it is easy to cause detail degradation, the removal effect of Gaussian noise and impulse noise in fuzzy text images is not good, and it is easy to lose the edge information of fuzzy text images

[0015] The third is that the existing technology has not grasped the core problem of text picture motion blur, has not analyzed the text picture, has not made statistics on its gradient distribution and brightness distribution, and has not found that the gradient distribution and brightness distribution of the text picture satisfy the prior knowledge Sparseness, the processing of fuzzy text images basically follows the idea of ​​ordinary image sharpening processing, there is no idea and method specifically for fuzzy text images, there is no targeted design combined with prior knowledge of fuzzy text images, and the blur kernel and restoration cannot be effectively estimated clear text images

[0016] Fourth, for the situation where the blur kernel is known, it cannot achieve the effect of high-quality restoration of text and pictures, and at the same time ensure the efficiency of the algorithm

There is no solution to the problem of artifacts in the recovered text images. The text image algorithm has high algorithm complexity and is not easy to implement. The quality of the image restoration effect is not stable and the difference is large. Blurred text image clarity processing control capabilities and services The quality assurance mechanism is poor, the judgment of the picture clearing processing effect is not accurate, and the picture clearing processing method is not robust and efficient

Images