Neigh Shrink image denoising method based on PCNN (Pulse Coupled Neural Network) region segmentation

Abstract

The invention belongs to the technical field of image processing, relating to a Neigh Shrink image denoising method based on PCNN (Pulse Coupled Neural Network) region segmentation. The Neigh Shrink image denoising method comprises the steps of: carrying out two-dimensional stable wavelet transformation on a noisy image f (x,y), and respectively acquiring subband coefficients: low-frequency coefficients, horizontal detail coefficients, vertical detail coefficients and diagonal detail coefficients; carrying out region segmentation on the first layer of the low-frequency coefficients by utilizing a PCNN, and recording image information obtained after segmentation as the domain PCNN; determining image neighborhood domain Neigh Shrink according to a Neigh Shrink method; setting dm and n as the image coefficients of the current threshold values, and obtaining the neighborhood required by the processing of the current threshold values by utilizing dm, n belongs to a set of the intersection of (domainPCNN) and (domainNeighShrink); keeping the low-frequency coefficients unchanged, and carrying out Neigh Shrink neighborhood threshold value processing respectively on each layer of the horizontal detail coefficients, the vertical detail coefficients and the diagonal detail coefficients obtained in the last step; and carrying out stable wavelet reconstruction on the low-frequency coefficients and filtered high-frequency subbands to obtain the denoised images. The invention can better restore original images, protect marginal information, and improve the denoising performance.

Description

technical field The invention belongs to the technical field of image processing, and in particular relates to a NeighShrink image denoising method based on PCNN region segmentation. Background technique Due to the influence of image acquisition equipment, image transmission process and storage devices, most digital images will be polluted by noise, and the image quality will be reduced, affecting visual effects and follow-up work such as image restoration, segmentation, feature extraction, and pattern recognition. Therefore, noise suppression is a very important task in image processing. Because wavelet transform has the characteristics of low entropy, multi-resolution, decorrelation and base selection flexibility, using wavelet transform to denoise images can achieve very good results. There are many methods for denoising by wavelet transform, among which the NeighShrink method [1, 2] sets a separate threshold for each coefficient and achieves a better image denoising ef...

AI Technical Summary

Problems solved by technology

However, when the NeighShrink method performs denoising processing, the edge information of the image will be suppressed excessively, or the neighborhood with stronger correlation will be artificially destroyed.