Image noise reduction method based on local Gaussian process regression

Abstract

The invention discloses an image denoising method based on local Gaussian process regression, and the method comprises the following steps: segmenting a noisy image into a plurality of overlapped image blocks, and for each pixel point, training a Gaussian process regression model by taking a neighborhood pixel containing local structure similarity information as a training sample set; constructing a composite covariance function to measure the similarity between input data, and predicting the corresponding output, namely a pixel value after noise reduction processing; and then, carrying out linear smoothing processing on overlapped regions of the image blocks, combining the processed image blocks in sequence, and reconstructing a noise-reduced image. Similarity information in a local structure of the image can be effectively utilized, structural information in an original image can be reserved while noise reduction is performed, the method has the advantages of being high in adaptability, robustness and reliability and the like, and effective noise reduction processing can be performed on the image polluted by Gaussian white noise.

Description

technical field [0001] The invention relates to the technical field of digital image processing, in particular to an image noise reduction method based on local Gaussian process regression. Background technique [0002] The 21st century is an era of informationization. Voice and text are no longer a simple form of information transmission, but have developed into multimedia forms including images, videos, and data. According to statistics, 70% of the information that humans receive from the outside world comes from images. With the rapid development of computer science and image processing technology, images have been widely used in various fields of human production and life, such as medical imaging, artificial intelligence, education and training, etc. However, digital images are inevitably affected by noise in the process of acquisition and transmission, which will destroy the structure and details of the image itself, and seriously affect the visual quality of the image...

AI Technical Summary

Problems solved by technology

However, this method takes the spatial coordinates of pixels and the corresponding pixel values ​​as training sample pairs, which cannot fully capture the correlation between neighboring pixels, and the denoising effect is average.

Images