Image denoising method based on decoupling depth dictionary learning

Abstract

The invention discloses an image denoising method based on decoupling depth dictionary learning, and the method comprises the steps: firstly carrying out the preprocessing of image data, and decomposing an input image into a structure image and a texture image; the structure image and the texture image respectively follow a dictionary learning model to learn a self-adaptive dictionary, a decoupling depth dictionary learning model is solved, and a model-driven network is utilized to fuse different components of the processed image; inputting the preprocessed image into a neural network model for parameter training to obtain a network model with a good denoising effect; and finally, inputting an image needing to be tested into the network model subjected to parameter training to obtain a clear image. The decoupling depth dictionary learning image denoising model formed by the method overcomes the problems of serious model degradation, slow network convergence, non-adaptive dictionary and the like in the current mainstream method, and effectively utilizes the learning ability of the DNN, so that the fine structure of the image can be well recovered in a serious noise environment.

Description

technical field [0001] The invention relates to a natural source image denoising method, in particular to an image denoising method based on decoupling deep dictionary learning, and belongs to the field of image denoising. Background technique [0002] With the rapid development of information science and technology, more and more research has been done on target detection, object recognition, image retrieval, etc. However, many applications such as target detection, object recognition, and image retrieval require the input of digital images as clear as possible. Digital images will be polluted by noise for various reasons in the process of acquisition and storage, so image denoising is a very important topic. [0003] At present, the existing denoising methods include: using K-SVD dictionary training algorithm for denoising, using nonlinear reaction diffusion algorithm (TNRD) for denoising, using median filtering for denoising, using wavelet transform for denoising and usin...

AI Technical Summary

Problems solved by technology

But this method learns a general dictionary to represent all images, reducing the flexibility of the model

There are also many technologies that integrate

Images