首页|基于改进群稀疏正则化的稀疏角度图像重建

基于改进群稀疏正则化的稀疏角度图像重建

扫码查看
原文链接
  • 万方数据
  • 维普
计算机断层扫描(Computer Tomography，CT)是临床医学中广泛使用的医学图像，可以清楚地显示人体的精细结构细节，为医生诊断疾病提供很大帮助。通过最近的研究表明，基于群稀疏正则化的联合代数重建技术(Simultaneous Algebraic Reconstruction Technique，SART)重建在稀疏角度采样背景下可以获得较好的性能。然而，群稀疏正则化在去掉伪影的同时，可能将边缘或细节过度平滑，使得对比度降低，无法获得符合人类视觉效果的高分辨率图像。因此，该文提出了一种基于改进群稀疏正则化的稀疏角度图像重建方法。首先对稀疏角度下的Shepp-Logan模型进行SART重建，再利用群稀疏正则化去除图像伪影，最后利用滚动引导滤波(Rolling Guided Filtering，RGF)进行对比度提升，再次作为SART的输入进行迭代。实验结果表明，该方法在视觉上以及PSNR(Peak Signal-to-Noise Ratio)，MSE(Mean Squared Error)和FSIM(Feature Similarity)上均优于其他算法，并且在迭代初始阶段就具有较好的收敛性能。
Sparse Angle Image Reconstruction Based on Improved Group-sparse Regularization
Computer Tomography(CT)is a widely used medical image in clinical medicine,which can clearly display the fine structural details of the human body,providing great help for doctors in diagnosing diseases.Recent research shows that the group sparse regularization based SART(Simultaneous Algebraic Reconstruction Technique)reconstruction can achieve better performance in the context of sparse angle sampling.However,while removing artifacts,group sparse regularization may over smooth the edges or details,re-ducing the contrast and making it impossible to obtain high-resolution images consistent with human visual effects.Therefore,we propose a sparse angle image reconstruction method based on improved group sparse regularization.Firstly,the Shepp-Logan model at the sparse angle is reconstructed by SART,and then the group sparse regularization is used to remove the image artifacts.Finally,the rolling guided filtering(RGF)is used to improve the contrast,and it is iterated again as the input of SART.The experimental results show that the proposed method outperforms other algorithms in terms of vision,PSNR(Peak Signal to Noise Ratio),MSE(Mean Squared Error),and FSIM(Feature Similarity),and has good convergence performance at the initial iteration stage.

sparse representationsimultaneous algebraic reconstruction techniquerolling guided filteringsparse angleimage reconstruc-tion

魏志晴、郑文康、白艳萍、谭秀辉、程蓉、胡红萍、王鹏

展开 >

中北大学 数学学院,山西 太原 030051

稀疏表示 联合代数重建技术 滚动引导滤波 稀疏角度 图像重建

山西省基础研究计划资助项目山西省基础研究计划资助项目国家重大科研仪器研制项目中北大学第18届研究生科技立项项目

2021030212241952021030212231896192780720221848

2024

10.3969/j.issn.1673-629X.2024.03.009

计算机技术与发展
陕西省计算机学会

计算机技术与发展

CSTPCD
影响因子：0.621
ISSN：1673-629X
年,卷(期)：2024.34(3)
  • 17