Salt and pepper noise removal method based on stationary Framelet transform with non-convex sparsity regularization DOI Creative Commons

Chen Ying-pin,

Yuming Huang, Lingzhi Wang

и другие.

arXiv (Cornell University), Год журнала: 2021, Номер unknown

Опубликована: Янв. 1, 2021

Salt and pepper noise removal is a common inverse problem in image processing. Traditional denoising methods have two limitations. First, characteristics are often not described accurately. For example, the location information ignored sparsity of salt by L1 norm, which cannot illustrate sparse variables clearly. Second, conventional separate contaminated into recovered part, thus resulting recovering an with unsatisfied smooth parts detail parts. In this study, we introduce detection strategy to determine position noise, non-convex regularization depicted Lp quasi-norm employed describe thereby addressing first limitation. The morphological component analysis framework stationary Framelet transform adopted decompose processed cartoon, texture, resolve second Then, alternating direction method multipliers (ADMM) solve proposed model. Finally, experiments conducted verify compare it some current state-of-the-art methods. experimental results show that can remove while preserving details image.

Язык: Английский

Salt and pepper noise removal method based on stationary Framelet transform with non‐convex sparsity regularization DOI Creative Commons
Yingpin Chen, Yuming Huang, Lingzhi Wang

и другие.

IET Image Processing, Год журнала: 2022, Номер 16(7), С. 1846 - 1865

Опубликована: Фев. 17, 2022

Abstract Salt and pepper noise occurs randomly causes image degradation. Numerous denoising methods have been proposed to suppress this noise. However, existing two main limitations. First, characteristics, such as location information sparsity, are often described inaccurately or even ignored. Second, many separate the contaminated into a recovered part, leading recovery of an with unsatisfactory smooth detailed parts. In study, authors introduce detection strategy determine position employ non‐convex sparsity regularization depicted by quasi‐norm describe noise, thereby addressing first limitation. We adopt morphological component analysis framework stationary Framelet transform decompose processed cartoon, texture, parts resolve second Then, model is applied using alternating direction method multipliers (ADMM). Finally, experiments conducted verify compare it some current state‐of‐the‐art methods. The experimental results show that can remove salt while preserving details outperforming

Язык: Английский

Процитировано

17
An L0 regularized cartoon-texture decomposition model for restoring images corrupted by blur and impulse noise DOI

Huasong Chen,

Zhenhua Xu,

Qiansheng Feng

и другие.

Signal Processing Image Communication, Год журнала: 2019, Номер 82, С. 115762 - 115762

Опубликована: Дек. 27, 2019

Язык: Английский

Процитировано

9
Robust sparse time‐frequency analysis for data missing scenarios DOI Creative Commons
Yingpin Chen, Yuming Huang, Jianhua Song

и другие.

IET Signal Processing, Год журнала: 2023, Номер 17(1)

Опубликована: Янв. 1, 2023

Sparse time-frequency analysis (STFA) can precisely achieve the spectrum of local truncated signal. However, when signal is disturbed by unexpected data loss, STFA cannot distinguish effective signals from missing interferences. To address this issue and establish a robust model for (TFA) in loss scenarios, stationary Framelet transform-based morphological component introduced STFA. In proposed model, processed regarded as sum cartoon, texture data-missing parts. The cartoon parts are reconstructed independently taking advantage transform. Then, forward-backwards splitting method employed to split into recovery imaging stages. two stages then solved separately using alternating direction multipliers (ADMM). Finally, several experiments conducted show performance under different levels, it compared with some existing state-of-the-art methods. results indicate that outperforms methods obtaining sparse missing. has potential value TFA scenarios where easily lost.

Язык: Английский

Процитировано

2
An L0-regularized global anisotropic gradient prior for single-image de-raining DOI

Huasong Chen,

Zhenhua Xu,

Yasong Zhang

и другие.

Applied Mathematical Modelling, Год журнала: 2021, Номер 98, С. 628 - 651

Опубликована: Июнь 18, 2021

Язык: Английский

Процитировано

5
Image Tone Mapping by Employing Anisotropic Total Variation and Two-Directional Gradient Prior DOI
Qi Zhang,

Huasong Chen,

Nina Hua

и другие.

Circuits Systems and Signal Processing, Год журнала: 2022, Номер 41(9), С. 5026 - 5048

Опубликована: Апрель 8, 2022

Язык: Английский

Процитировано

3
Single image deraining using local rain distribution map DOI

Huasong Chen,

Jing Wu,

Zhenhua Xu

и другие.

Multimedia Tools and Applications, Год журнала: 2023, Номер 83(17), С. 50349 - 50380

Опубликована: Ноя. 6, 2023

Язык: Английский

Процитировано

0
MRI Reconstruction using Minimax-Concave Total Variation Regularization based on p-norm DOI
Yongxu Liu, Xiaoyan Fu, Yu Song

и другие.

2022 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Год журнала: 2022, Номер unknown, С. 1206 - 1212

Опубликована: Окт. 9, 2022

Magnetic resonance imaging (MRI) reconstruction model based on total variation (TV) regularization can solve some problems, e.g., incomplete reconstruction, blurred imaging, and denoising. However, it has problems such as sensitivity to outliers, poor ability induce the sparsity of gradient domain MR image. In this paper, minimax-concave $L_{p}-$norm (MCTV-L p ) is proposed overcome these drawbacks. Specifically, TV-L constructed using exponent ${p}(0\lt{p}\lt 1)$, which defined gradient. Then combined with penalty construct MCTV-L . Finally, sparse (MCTV-SRM) proposed, where objective function formulated sum data-fitting term $L_{2}-$norm. Moreover, an optimization algorithm alternating direction method multipliers (ADMM) given related iteratively. Results different datasets experimental settings show that better adapted MRI relative error PSNR are significantly improved than several typical methods, while reconstruct images clear details textures.

Язык: Английский

Процитировано

0
Exponential-Ant Cuckoo Search Optimization for image deblurring with spinal cord images based on kernel estimation DOI

S. Shanmuga Priya,

S. Letitia

Signal Image and Video Processing, Год журнала: 2021, Номер 16(2), С. 339 - 347

Опубликована: Июнь 22, 2021

Язык: Английский

Процитировано

0
Salt and pepper noise removal method based on stationary Framelet transform with non-convex sparsity regularization DOI Creative Commons

Chen Ying-pin,

Yuming Huang, Lingzhi Wang

и другие.

arXiv (Cornell University), Год журнала: 2021, Номер unknown

Опубликована: Янв. 1, 2021

Salt and pepper noise removal is a common inverse problem in image processing. Traditional denoising methods have two limitations. First, characteristics are often not described accurately. For example, the location information ignored sparsity of salt by L1 norm, which cannot illustrate sparse variables clearly. Second, conventional separate contaminated into recovered part, thus resulting recovering an with unsatisfied smooth parts detail parts. In this study, we introduce detection strategy to determine position noise, non-convex regularization depicted Lp quasi-norm employed describe thereby addressing first limitation. The morphological component analysis framework stationary Framelet transform adopted decompose processed cartoon, texture, resolve second Then, alternating direction method multipliers (ADMM) solve proposed model. Finally, experiments conducted verify compare it some current state-of-the-art methods. experimental results show that can remove while preserving details image.

Язык: Английский

Процитировано

0