%0 Journal Article
%T 基于NSCT稀疏编码的医学图像超分辨率重建算法<br>Super-Resolution Reconstruction of Medical Images Based on Sparse Representation in the NSCT Domain
%A 聂小栋
%A 成云凤
%J Software Engineering and Applications
%P 276-292
%@ 2325-2278
%D 2023
%I Hans Publishing
%R 10.12677/SEA.2023.122028
%X 为了满足临床对医学图像分辨率的进一步需求，提出了一种非下采样Contourlet变换和稀疏表示结合的单帧医学图像超分辨率重建算法。该方法在改进字典学习步骤中，首先通过非下采样Contourlet变换中的方向滤波器提取训练样本集图像的高频特征；其次，采用K-SVD奇异值分解法和OMP正交匹配追踪算法来求取高、低分辨率字典和稀疏表示系数，并重建出高分辨率图像。最后，将提出的方法与Yang、Zeyde提出的稀疏表示重建算法都应用于医学图像超分辨率重建上，通过实验对比分析，提出算法重建结果的PSNR值与SSIM值分别达到45.389和0.907，充分展现出了提出算法的优势。<br />
In order to meet the additional demands of image resolution in clinical medicine, a super-resolution reconstruction algorithm based on sparse representation in the nonsubsampled contourlet transform (NSCT) domain is proposed. In this method, in the improved dictionary learning step, firstly, the high-frequency features of the training sample set images are extracted by the direction filter in the non-downsampling Contourlet transform; secondly, the K-SVD singular value decomposition method and the OMP orthogonal matching tracking algorithm are used to find the high- and low-resolution dictionaries and sparse representation coefficients, and reconstruct the high-resolution images. Finally, the proposed method and the sparse representation reconstruction algorithm proposed by Yang and Zeyde are both applied to the super-resolution reconstruction of medical images. Through experimental comparison and analysis, the PSNR and SSIM values of the reconstruction results of the proposed algorithm reach 45.389 and 0.907 respectively, fully demonstrating the advantages of the proposed algorithm.
%K 医学图像超分辨率重建，非下采样Contourlet变换，字典学习，稀疏表示<br>Medical Image
%K Super-Resolution Reconstruction
%K Nonsubsampling Contourlet Transform
%K Dictionary Learning
%K Sparse Representation
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=64315