Image noise estimation method based on human eye visual features and partitioning analysis method

Abstract

An image noise estimation method based on human eye visual features and a partitioning analysis method comprises the steps of utilizing a human eye contrast sensitivity function for processing an original image with noise to obtain an initially-processed image; utilizing a watershed partitioning algorithm for conducting approximation region partitioning on the initially-processed image to obtain a plurality of partitioned image region blocks, and obtaining a region partitioned image; conducting noise-free image approximation rebuilding on partitioned regions of the region portioned image to obtain a rebuilt estimated noise-free image of the whole image; according to the original image with the noise and the rebuilt estimated noise-free image, obtaining a distribution diagram of intensity-noise pairs, and utilizing the distribution diagram of the intensity-noise pairs for obtaining a noise label of the original image with the noise. According to the image noise estimation method, due to the combination of the human eye visual features, partitioning analysis and noise estimation are carried out on the original observation image, a single comprehensive estimated label value is obtained finally, and the result is quite approximate to a human eye visual system.

Description

technical field [0001] The invention relates to image processing technology, in particular to an image noise estimation method based on human visual features and block analysis method. Background technique [0002] With the development of human society towards a high degree of digitization, the rapid development and popularization of digital images, digital video and digital television will also become inevitable. In the various technologies of digital image processing, digital images may be subject to various degradation distortions in the process of acquisition, compression, storage, transmission and reconstruction, especially noise, which will inevitably lead to the problem of image degradation. The problem of how to evaluate image noise more effectively has also emerged and has become a research hotspot in image processing. [0003] Because the image is ultimately viewed by a human, the best way to evaluate noise is the subjective evaluation of the human eye. However, ...

AI Technical Summary

Problems solved by technology

However, these algorithms are difficult to deal with certain blurred or image content-sensitive images, resulting in limited application

[0005] In addition, most noise evaluation methods are accurate and strict in definition, simple and easy to implement, and can better determine the noise level difference between images, but generally do not consider the visual psychological factors of image observers, and the subject of image evaluation— —People often play a very important role in image evaluation, so the evaluation results of objective evaluation methods often cannot match the results of subjective evaluation by human eyes

Images