Image noise variance estimation method based on integer DCT

Abstract

The invention discloses an image noise variance estimation method based on integer DCT. According to resolution of a noised image to be estimated, image blocks which are of certain sizes and are not overlapped with one another are obtained through division, transformation coefficients are obtained by performing the integer DCT on the divided image blocks, by taking a transformation AC coefficient as a tolerance of image uniformity, selecting most uniform blocks from obtained uniform blocks according to a certain proportion, and a reference variance of image noise is calculated; according to the reference variance of the image noise, re-searching the uniform blocks; and re-obtaining the uniform blocks, and a noise variance of the noised image to be estimated is obtained by performing weighting processing according to uniformity of the blocks. According to the invention, the integer DCT facilitates hardware realization, the method is not restricted to noised images with specific features, at the same time, through correcting search intervals of the uniform blocks by use of a DC coefficient, an under-estimation phenomenon generated due to noise inhibition caused by too bright or too dark images can be improved, and the image noise estimation method provided by the invention has the advantages of high estimation accuracy and high robustness.

Description

Technical field [0001] The invention belongs to the field of digital image processing, and in particular relates to an image noise estimation method based on integer DCT transform. Background technique [0002] Image is a vivid, specific and effective information transmission medium. It is inevitable to introduce noise in the process of image acquisition and transmission. In order to restore the original real information of the image as much as possible, image denoising becomes the basis of other image preprocessing. In many image denoising methods, estimating the noise function model and its related parameters is one of the key tasks of image denoising. The noise model generally used is the Gauss model, because Gauss function noise is the most common in image noise, and noise variance is an important indicator for evaluating noise. [0003] There are three types of commonly used image noise variance estimation methods: (1) Filtering method. Based on smoothing filtering process...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an image noise estimation method based on integer DCT transformation, aiming at solving the problem of underestimation in the current image noise variance estimation method when the image is too bright or too dark, and the calculation is complicated and not conducive to hardware implementation. The block method is highly dependent on the image's own characteristics and lacks robustness

Images