Image noise estimation method and device and storage medium

Abstract

The invention relates to the technical field of image processing, and discloses an image noise estimation method and device and a storage medium. According to the method, a to-be-estimated image is partitioned, the obtained noisy image blocks are screened to obtain the to-be-estimated noisy image blocks, and the corresponding number of new noisy image blocks are obtained by adding a plurality of Gaussian white noises with different noise standard deviations into the to-be-estimated noisy image blocks, then the noise standard deviation estimation value of the to-be-estimated noise-containing image block is calculated through the mean value of the tail singular values of the new noise-containing image blocks, and the noise standard deviation estimation values of the to-be-estimated noise-containing image blocks are averaged to obtain a noise standard deviation estimation value of the to-be-estimated image; according to the embodiment of the invention, the estimation of the original noise standard deviation is realized by adding random noise for multiple times, so that the contingency of the estimation method is reduced in probability, the estimation result is more stable, and the estimation precision is improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image noise estimation method, device and storage medium. Background technique [0002] Noise is inevitable in the process of image acquisition, transmission and extraction, and these noises will seriously affect the quality of the image, so it is necessary to estimate the noise of the image. [0003] When performing noise estimation in the prior art, random noise of known fixed intensity is first added to the noisy image to form a new noisy image, and then the relationship between the tail singular value mean of the original noisy image and the new noisy image is used Implements the estimation of the noise standard deviation. [0004] Due to the randomness of the Gaussian noise in the original noisy image, adding random noise to the noisy image only once will result in a large randomness, which makes the estimation result unstable and low in accuracy. Cont...

AI Technical Summary

Problems solved by technology

[0005] The invention provides an image noise estimation method, device and storage medium, which solves the technical problems of poor stability and low estimation accuracy of existing noise estimation methods

Images