A Gaussian-Pulse Hybrid Image Noise Removal Method

Abstract

The invention discloses a Gaussian-pulse mixed image noise removal method, which includes detecting the position of random pulse noise on a target image and filtering to obtain a filtered image, and recording the position of random pulse noise with a probability matrix; The final image is decomposed into multiple image blocks, and the corresponding image sub-blocks are found on the target image to be stacked into a tensor form and the corresponding tensor low-rank-sparse decomposition model is established; each tensor low-rank-sparse decomposition model is solved The optimization problem in , and use the alternating direction method to achieve decomposition and iterative optimization until convergence; obtain the low-rank tensor of each tensor low-rank-sparse decomposition model and expand it into a matrix form, and obtain the denoising result of each image sub-block and The overlapping area is averaged to obtain the final denoising result. The implementation of the present invention effectively fuses the local and non-local statistical characteristics of the image, overcomes the problem of insufficient detection accuracy of the adaptive median filter, and simultaneously filters Gaussian-impulse mixed noise to achieve a denoising effect with a higher signal-to-noise ratio.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a Gaussian-pulse mixed image noise removal method. Background technique [0002] During the acquisition and transmission of image signals, the quality is often degraded due to the interference of various external noises, which seriously affects the subsequent processing of images, such as edge detection, target recognition, feature extraction, image segmentation, etc. Therefore, image denoising has become an image The most basic and important link in the processing process, and has attracted widespread attention. However, most of the existing image denoising methods are oriented towards zero-mean Gaussian noise, such as traditional anisotropic diffusion, nonlinear diffusion, bilateral filtering, non-local mean and their related denoising methods. In fact, in practical applications, image noise is diverse. Some scholars believe that the noise generated in the image imagi...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the embodiment of the present invention is to provide a Gaussian-pulse mixed image noise removal method, which can face noise points between 0-255, and overcome the problem of insufficient accuracy of the adaptive median filter when detecting pulse noise , to effectively fuse the local and non-local statistical characteristics of the image to achieve a denoising effect with a higher signal-to-noise ratio

Images