Deep neural network image denoising method based on global smooth constraint prior model

Abstract

The invention discloses a deep neural network image denoising method based on a global smooth constraint prior model. The method mainly comprises the following steps: constructing global smoothness constraint priori by using non-local smoothness of a natural image; constructing an image denoising cost function solved in a transform domain according to the global smooth constraint priori obtained in the previous step; optimizing the cost function constructed in the previous step by using a gradient-based method to obtain an image denoising iteration framework based on a global smooth constraint prior model; expanding the iteration framework into a deep neural network model; training the constructed deep neural network model; and taking a noise image as input, and obtaining a recovered high-quality image by using the deep neural network model trained in the previous step. According to the method, a good denoising effect can be obtained, and the method is an effective image denoising method.

Description

technical field [0001] The invention relates to image denoising technology, in particular to a deep neural network image denoising method based on a global smoothness constraint prior model, and belongs to the field of digital image processing. Background technique [0002] Image denoising is a widely researched hot topic in the field of digital image processing, which has very important practical application value. Its purpose is to restore high-quality image x from noisy image y, improve image quality, and provide further analysis And lay the foundation for processing. The degradation model can be expressed as y=x+v, where v is generally assumed to be additive white Gaussian noise (AWGN) with standard deviation σ. Image denoising is a typical ill-conditioned inverse problem. Therefore, how to use the prior knowledge of the image to constrain the solution process becomes particularly critical. [0003] Image denoising methods can generally be divided into two categories:...

AI Technical Summary

Problems solved by technology

Compared with most model-based methods, their reconstruction speed is faster, and they can recover better the texture and structure of the image, but they are also prone to false texture information.

Images