Adaptive orthogonal wavelet image denoising method based on accurate local variance priori modeling

Abstract

The invention provides an adaptive orthogonal wavelet image denoising method based on accurate local variance priori modeling. The method is characterized by carrying out accurate modeling on orthogonal wavelet transform detail coefficient local variance prior distribution through maximum likelihood estimation, thereby realizing a better denoising performance, improving peak signal-to-noise ratio of a de-noised image, and improving visual effect of the de-noised image. The beneficial effects are that the method overcomes the defect that an existing method is not high in local variance estimation, can better express statistical property of the image orthogonal wavelet decomposition detail coefficient, can give a denoising result adaptively, can remove additive white Gaussian noise in a natural image more effectively, and meanwhile, can better protect regions having rich edge and texture detail information and the like in an original image, and improves visual effect and peak signal-to-noise ratio of the de-noised image. The method is low in computation complexity, and is suitable for immense image denoising application in the age of big data.

Description

technical field [0001] The invention relates to an adaptive orthogonal wavelet image denoising method based on accurate local variance prior modeling. Background technique [0002] At present, noise is unavoidable in the process of image acquisition and transmission, and image denoising is an important research topic in the field of image processing. Wavelet transform is an efficient multi-resolution time-frequency analysis method. Orthogonal wavelet transform has the advantages of high computational efficiency and small coefficient redundancy, and is widely used in the field of image denoising. Under the denoising framework of additive Gaussian white noise, given the wavelet transform coefficient y=x+n of the noise image obtained, the purpose of denoising is to restore the original clean image x without distortion as much as possible while suppressing the noise n . [0003] The basic principle of wavelet denoising is to use the different characteristics of image informati...

AI Technical Summary

Problems solved by technology

[0004] However, due to the poor estimation of the exponential prior parameter λ in its method, the modeling effect on the local variance θ is not ideal, which directly affects the performance of the denoising algorithm

The peak signal-to-noise ratio of the denoised image is not high, and the areas with rich details such as edges and textures in the denoised image have serious artifacts and poor visual effects

Images