Image denoising method based on noise estimation

Abstract

The invention discloses an image denoising method based on noise estimation, which is characterized in that an image is divided into a plurality of homologous regions according to the image content by using super-pixel segmentation, the prior knowledge that flat information can better represent the noise pollution level of the image is utilized, smooth homologous regions in the image are found by taking an image information entropy, the noise standard deviation of the smooth regions is estimated, and the noise standard deviation of the smooth regions is taken as the noise level of the whole image, so that a purpose of more accurately estimating the noise level is achieved, thus a nonlocal mean (NLM) denoising method is modified according to the noise level, the denoising degree is reasonably controlled according to the noise degree, the image with noise can be processed adaptively, the overall effect of the denoised image is enabled to be greatly improved than that of a traditional NLM algorithm, and details are better reserved while denoising. The whole process can realize automation and self-energy and does need manual intervention.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image denoising method for noise estimation. Background technique [0002] With the rapid development of digital image and multimedia technology, there are more and more various types of optical imaging systems, and people have higher and higher requirements for the picture quality of imaging systems. This kind of noise pollution will reduce the quality of the acquired image and the original image, which will not only affect the effect of the image, but also affect the readability of the image when the noise is serious, resulting in errors in subsequent information acquisition. Therefore, in the field of image processing, it is very meaningful to denoise digital images. [0003] At present, a commonly used denoising method is the non-local mean (NLM) image denoising algorithm with excellent performance and strong edge preservation ability. The basic idea is to find pixels sim...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to propose an adaptive, easy-to-implement, and robust image denoising method, so as to solve the problem of insufficient denoising degree or excessive denoising caused by the unknown noise level of existing image denoising methods , and effectively preserve edge and texture information while denoising

Images