Two-dimension analysis thinning model and dictionary training method and image denoising method thereof

Abstract

The invention discloses a two-dimension analysis thinning model and a dictionary training method and an image denoising method based on the two-dimension analysis thinning model, wherein the two-dimension analysis thinning model fully utilizes the spatial correlation of images, needs fewer training samples, and greatly saves the storage space of a dictionary.

Description

technical field [0001] The invention belongs to the technical field of signal modeling, and in particular relates to a two-dimensional analysis sparse model, a dictionary training method and an image denoising method based on the model. Background technique [0002] Sparse representation is a relatively mature modeling method that has been widely studied and widely used in most signal processing fields, such as image denoising, texture synthesis, video processing and image classification. Using sparse representation to signal Modeling mainly includes two categories: synthetic sparse modeling and analytical sparse modeling. [0003] The synthetic model is defined as follows: x=Db,s.t.||b|| 0 = k, here is an overcomplete dictionary in which each column represents an atom (primitive). is a sparse vector. l 0 Norm||·|| 0 The sparsity used to characterize the sparse signal is defined as k non-zero elements in a vector. In the synthetic model, the signal x can be repres...

AI Technical Summary

Problems solved by technology

In the past, when all the analysis sparse models dealt with two-dimensional images, they converted them into one-dimensional high-dimensional signals by columns or rows, which caused the following problems: first, the two-dimensional spatial structure of the image was destroyed, and the local Correlation is not exploited effectively, and secondly, in order to get an effective robust estimate, a large number of training samples in high-dimensional space are required

Images