De-noising method based on external block autoencoding learning and internal block clustering

Abstract

The invention relates to a de-noising method based on external block autoencoding learning and internal block clustering. The method comprises learning block structure features from an external clean natural image block by using an autoencoding model in deep learning, reducing dimensions of a noise image by using the features, achieving block clustering within a whole image range by using a strategy from coarse to fine, constructing a lowrank regular constraint in each class, constructing a global constraint in all classes, establishing a total energy function, and de-noising the target image by means of energy minimization. The method assists internal block clustering de-noising of an image to be tested by using the external natural image block structure information, and solves a problem that a conventional de-noising method is not good in de-noising effect on natural images corroded by Gaussian white noise.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a denoising method based on outer block self-encoding learning and inner block clustering. Background technique [0002] The purpose of image denoising is to restore a potential clean image from the noise map, which is essentially the inverse process of image degradation. As an important low-level vision problem, image denoising techniques have a long history. Early image denoising methods usually use linear or nonlinear local smoothing filtering, such as Gaussian filtering and heterogeneous diffusion. However, most of these edge-preserving algorithms ignore the global information of the image, and it is difficult to achieve the highest level of image denoising. In recent years, a non-local averaging (NLM) method based on the internal self-similarity of images has received a lot of attention. On this basis, many successful denoising algorithms based on image blocks hav...

AI Technical Summary

Problems solved by technology

Since the full search is computationally expensive, block matching is usually limited to a small window in the image

In addition, some important texture edges and corners in natural images do not have repeated structures locally, so excessive forcing of local self-similarity will lead to false edges and affect the visual quality of images

Images