Non-Blind图像反卷积论文整理
1 Spatial Deconvolution
Stochastic Deconvolution 2013
http://www.cs.ubc.ca/labs/imager/tr/2013/StochasticDeconvolution/
在空域上直接逐点探测图像灰度数据应该增加,还是减少,并把结果应用到该点。判断的依据就是变化后的值|Ax - y| + w|Dx| 谁更小。如此迭代多次,数据收敛。
其优点是卷积因子可在空域上变化;非常直观,方便设计更好的先验知识模型。
缺点是速度很慢。
Inverse Kernels for Fast Spatial Deconvolution 2014
http://lxu.me/
空域上进行快速反卷积
迭代过程中,通过augmented Lagrangian method得到固定的反卷积的空域因子 kernel,再使用SVD 方法对kernel分解成多个一维的卷积因子加速空域卷积处理
2 Machine Learning
Deep Convolutional Neural Network for Image Deconvolution 2014
http://lxu.me/
A machine learning approach for non-blind image deconvolution 2013
http://webdav.is.mpg.de/pixel/neural_deconvolution/
3 TOTAL VARIATION
A NEW ALTERNATING MINIMIZATION ALGORITHM FOR TOTAL VARIATION IMAGE RECONSTRUCTION 2008
http://www.caam.rice.edu/~yzhang/reports/tr0710_rev.pdf
http://www.caam.rice.edu/~optimization/L1/ftvd/
http://www.caam.rice.edu/~optimization/L1/
L1-Related Optimization Software
FPC: MATLAB Algorithms for Large-Scale L1-Minimization: - Link
FTVd: A MATLAB Code for Total Variation based Image Deconvolution: - Link
Bregman Iterative Algorithms for Compressed Sensing: - Link
http://www.caam.rice.edu/~heinken/software/matlab_impl_constr/implicit_constraints.html
4 Constraints
Handling Outliers in Non-blind Image Deconvolution 2011
http://cg.postech.ac.kr/research/deconv_outliers/
Fast Deconvolution with Color Constraints on Gradients 2012
http://vision.seas.harvard.edu/ccap/
Fast High-Quality non-Blind Deconvolution Using Sparse Adaptive Priors 2014
http://www.inf.ufrgs.br/~oliveira/pubs_files/pubs.html
Good Image Priors for Non-blind Deconvolution: Generic vs Specific 2014
http://cs.brown.edu/~lbsun/GoodPriors2014/goodpriors2014eccv.html
Discriminative Non-blind Deblurring 2013
5 优化
Image Deconvolution Ringing Artifact Detection and Removal via PSF Frequency Analysis 2014
http://www.groupes.polymtl.ca/amosleh/papers/ECCV14.pdf
6 AUGMENTED LANGRANGIAN
IMAGE DEBLURRING BY AUGMENTED LANGRANGIAN WITH BM3D FRAME PRIOR 2010
http://www.cs.tut.fi/~foi/GCF-BM3D/
Inverse Kernels for Fast Spatial Deconvolution 2014
见上Spatial Deconvolution
7 Primal-Dual
A Primal-Dual Projected Gradient Algorithm for Efficient Beltrami Regularization
http://www.math.ucla.edu/~zosso/code.html
8 Half-Quadratic Regularization
Fast Image Deconvolution using H