An Image Denoising Method Based on Noise Prediction

Abstract

The invention discloses an image denoising method based on noise prediction, which belongs to the technical field of image processing. The present invention first removes unnecessary redundant information in the original image through the KSVD algorithm, and then performs non-mean value processing on the remaining image redundant information after the KSVD algorithm processing through the neighborhood window, avoiding inaccuracy caused by large-scale prediction noise, The image after removing the predicted noise can retain the detail information to a greater extent, and then use the difference between the original image and the non-mean processed image as the observed image of the independent component analysis algorithm, and ensure the separation of useful information components and noise through independent component analysis The components are independent of each other, and the useful information of the image is preserved as much as possible while denoising, so as to ensure the optimal image after denoising.

Description

technical field [0001] The invention discloses an image denoising method based on noise prediction, which belongs to the technical field of image processing. Background technique [0002] With the development of computer network-based information technology, people have higher and higher requirements for digital image quality, and the thermal noise, shot noise, and quantum noise generated by image output equipment and some optoelectronic devices will seriously reduce the final image quality. The quality of the output image. Digital picture noise can be divided into internal noise and external noise according to the cause, and can be divided into Gaussian noise and Rayleigh noise according to the shape of the noise amplitude distribution over time. Gaussian noise and salt-and-pepper noise are two common types in digital images. noise. [0003] The traditional denoising method treats noise and useful information equally, and thinks that noise and image detail information are...

AI Technical Summary

Problems solved by technology

[0003] The traditional denoising method treats noise and useful information equally, and thinks that noise and image detail information are distributed in different intervals in a certain transform domain. However, noise and some useful information images are generally superimposed on each other in a certain area. While removing noise, it will also remove useful detail information. Different denoising methods can only be effective within a specific noise signal range. Noisy images that exceed the range will be distorted in large areas after processing.

Images