Blind image deblurring method based on l0 regularization and blur kernel post-processing

Abstract

The invention discloses a blind image deblurring method based on L0 regularization and blur kernel post-processing, which introduces prior information about image gradients, blur kernel pixels and blur kernel gradient sparsity into the optimal model of image restoration, and It is expressed in the form of L0 regular term; secondly, the fuzzy kernel obtained by the optimization calculation is post-processed according to its objective characteristics, and human intervention is used to make up for the shortcomings caused by the optimization model, so that the restored fuzzy kernel and intermediate image are more consistent with reality. The quality of the final restored image is further improved; finally, the semi-quadratic splitting method is used to solve the optimal model. The solution is simple and reduces the amount of calculation. At the same time, it is combined with the pyramid model for hierarchical calculation, so the present invention has high robustness and a wide range of applications.

Description

technical field [0001] The present invention relates to the technical field of image restoration, and more specifically, to a method for blind image deblurring based on L0 regularization and blur kernel post-processing. Background technique [0002] With the development of society, images have become an important way of information dissemination and acquisition. However, due to the limitations of the imaging system, factors such as dust in the air, light, and weather during the imaging process will have a negative impact on the quality of the image, and image quality degradation is common. At the same time, the degradation of image quality will cause the loss of a large amount of information, and the image with degraded quality is very inconvenient in actual use, and even cannot be used directly. Therefore, it is of great significance to recover clear and high-quality images from degraded images. [0003] Blind image deblurring (blind image deconvolution) is one of the imp...

AI Technical Summary

Problems solved by technology

In the existing technical methods, the L2 regularization term of the fuzzy kernel is usually introduced into the optimization model. This method can quickly solve the problem, but the obtained fuzzy kernel is relatively dense, which does not meet the objective characteristics of the sparseness of the fuzzy kernel. The final restored poor image quality

There are also some methods that add the L1 regularization term of the blur kernel to the optimization model, but this will make the blur kernel contain more noise, and the image restoration effect is relatively poor

Images