该算法首先通过字典学习得到含噪图像的冗余字典，然后对相似的图像块进行聚类构成块群，并通过迭代收缩和L1正则化约束，对同类的图像块在字典上进行稀疏表示，以达到降噪的目的。实验结果表明，在常规的图像处理上，本文提出的算法能较好的保留图像的结构信息，与K-SVD和BM3D等现有的流行算法相比，具有更高的峰值信噪比（PSNR）
资源文件列表
CSR_Denoising/Data/images/barbara.tif , 264168
CSR_Denoising/Data/images/boat.tif , 264168
CSR_Denoising/Data/images/cameraman.tif , 65722
CSR_Denoising/Data/images/couple.tif , 264168
CSR_Denoising/Data/images/fingerprint.tif , 264168
CSR_Denoising/Data/images/hill.tif , 264168
CSR_Denoising/Data/images/house.tif , 67368
CSR_Denoising/Data/images/Lena512.tif , 264168
CSR_Denoising/Data/images/man.tif , 264168
CSR_Denoising/Data/images/Monarch_full.tif , 66196
CSR_Denoising/Data/images/peppers256.tif , 67368
CSR_Denoising/Data/images/straw.tif , 67368
CSR_Denoising/Data/images/textures/hex.tif , 58224
CSR_Denoising/Denoising_Main.m , 954
CSR_Denoising/Results/Denoising_results/nsig_20/CSR_de_PSNR_SSIM.txt , 0
CSR_Denoising/Results/Denoising_results/nsig_20/PCA_CSR_PSNR_SSIM.txt , 0
CSR_Denoising/Utilities/Add_noise.m , 191
CSR_Denoising/Utilities/cal_ssim.m , 6372
CSR_Denoising/Utilities/Clustering_PCA_New.m , 4373
CSR_Denoising/Utilities/csnr.m , 501
CSR_Denoising/Utilities/CSR_Denoising.m , 4942
CSR_Denoising/Utilities/CSR_Thresholding.m , 2648
CSR_Denoising/Utilities/ext_im.m , 356
CSR_Denoising/Utilities/find_blks.m , 1793
CSR_Denoising/Utilities/find_blks_fast.m , 1905
CSR_Denoising/Utilities/find_blks_fast2.m , 2966
CSR_Denoising/Utilities/getpca.m , 442
CSR_Denoising/Utilities/Image_Denoising.m , 1618
CSR_Denoising/Utilities/Proc_cls_idx.m , 192
CSR_Denoising/Utilities/Set_PCA_idx.m , 1247
CSR_Denoising/Utilities/soft.m , 121
CSR_Denoising/Utilities/Soft_PCA.m , 2542