文件列表:

Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\cellnorm2D.m (613, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\CoeffOper2D.m (1278, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\ConvSymAsym2D.m (1696, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\FraDec2D.m (835, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\FraDecMultiLevel2D.m (1135, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\FraRec2D.m (897, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\FraRecMultiLevel2D.m (1377, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\2D\GenerateFrameletFilter.m (778, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\aircraft.jpg (3590, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\slope.png (11366, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\2DTWFT\Usage.m (1578, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\bid_rgtv_c2f_cg.p (1088, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Copy_Enlarge_h.p (251, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Deblur_GL_CG_4.p (733, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Deconvolution_FHLP.p (330, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Fast_Image_Deconvolution_using_Hyper-Laplacian_Priors\fast_deconv.m (4634, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Fast_Image_Deconvolution_using_Hyper-Laplacian_Priors\kernels.mat (2175, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Fast_Image_Deconvolution_using_Hyper-Laplacian_Priors\snr.m (678, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Fast_Image_Deconvolution_using_Hyper-Laplacian_Priors\solve_image.m (6101, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\Fast_Image_Deconvolution_using_Hyper-Laplacian_Priors\test_fast_deconv.m (1914, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\fftconv.p (376, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\G_padding.p (160, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\informative_edge_mask_adaptive_mine.p (535, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\kernel_centralize.p (455, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\kernel_filter.p (193, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\kernel_solver_L2.p (854, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\k_rescale.p (130, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\sort_filter.p (306, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\TV_denoising.p (920, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\weights_computation.p (382, 2019-01-10)
Graph_Based_BID\Graph_Based_BID_p1.1\weight_function_l1.p (143, 2019-01-10)
Graph_Based_BID\graph_blind_main.m (1340, 2019-01-10)
Graph_Based_BID\Testing_Samples\10_1_blurred.png (363649, 2019-01-10)
Graph_Based_BID\Testing_Samples\1_2_blurred.png (441665, 2019-01-10)
Graph_Based_BID\Testing_Samples\35_1_blurred.png (385244, 2019-01-10)
Graph_Based_BID\Testing_Samples\39_6_blurred.png (350569, 2019-01-10)
Graph_Based_BID\Testing_Samples\44_4_blurred.png (340084, 2019-01-10)
Graph_Based_BID\Testing_Samples\4_4_blurred.png (355200, 2019-01-10)
Graph_Based_BID\Testing_Samples\building_blur.png (637714, 2019-01-10)
... ...

# Graph-Based-Blind-Image-Deblurring This code is the upgraded implementation of our TIP paper "Graph-based Blind Image Deblurring from a Single Photograph". ## Prerequisite Matlab(>=R2015a) ## Running the tests ``` Step 1. run graph_blind_main.m Step 2. select a blurred image ``` ## Parameters Users only need to tune *ONE* parameter. On line 21, the estimated kernel size ***k_estimate_size***. * The ***k_estimate_size*** must be *LARGER* than the real kernel size (The default value is 69). * In order to have the best performance, please set the value close to real kernel size and slightly larger. If you want to turn off the intermediate output, you can set *show_intermediate=false* on line 22. ## About noise In order to be more robust with noise, we add several denoising modules beyond the paper. * We embed a TV denoising to pre-process the input image. * We add a wavelet domain filtering for intermediate output kernels. * We add a mask to filter small/noisy gradient in the gradient domain. More sophisticated denoising, such as BM3D, can be done by users in advance. ## About Non-blind image deblurring After kernel estimation with the proposed algorithm, we use the state-of-the-art methods to do non-blind image deblurring. Here, we provide users with [1] to do the following non-blind image deblurring process. Users can also employ [2] or the non-blind deblurring method in [3], by themselves. [1] D. Krishnan and R. Fergus, “Fast image deconvolution using hyperlaplacian priors,” in Proceedings of Neural Information Processing Systems, 2009, Conference Proceedings, pp. 1033–1041. [2] D. Zoran and Y. Weiss, “From learning models of natural image patches to whole image restoration,” in Proceedings of IEEE International Conference on Computer Vision, 2011, Conference Proceedings, pp. 479–486. [3] J. Pan, D. Sun, H. Pfister, and M.-H. Yang, “Blind image deblurring using dark channel prior,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, June 2016. ## Citation ``` @ARTICLE{GraphBID, author={Y. Bai and G. Cheung and X. Liu and W. Gao}, journal={IEEE Transactions on Image Processing}, title={Graph-Based Blind Image Deblurring From a Single Photograph}, year={2019}, volume={28}, number={3}, pages={1404-1418}, doi={10.1109/TIP.2018.2874290}, ISSN={1057-7149}, month={March},} ```

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：