×

Variance-stabilization-based compressive inversion under Poisson or Poisson-Gaussian noise with analytical bounds. (English) Zbl 1429.94028

Summary: Most existing bounds for signal reconstruction from compressive measurements make the assumption of additive signal-independent noise. However in many compressive imaging systems, the noise statistics are more accurately represented by Poisson or Poisson-Gaussian noise models. In this paper, we derive upper bounds for signal reconstruction error from compressive measurements which are corrupted by Poisson or Poisson-Gaussian noise. The features of our bounds are as follows: (1) the bounds are derived for a computationally tractable convex estimator with statistically motivated parameter selection. The estimator penalizes signal sparsity subject to a constraint that imposes a novel statistically motivated upper bound on a term based on variance stabilization transforms to approximate the Poisson or Poisson-Gaussian distributions by distributions with (nearly) constant variance. (2) The bounds are applicable to signals that are sparse as well as compressible in any orthonormal basis, and are derived for compressive systems obeying realistic constraints such as non-negativity and flux-preservation. Our bounds are motivated by several properties of the variance stabilization transforms that we develop and analyze. We present extensive numerical results for signal reconstruction under varying number of measurements and varying signal intensity levels. Ours is the first piece of work to derive bounds on compressive inversion for the Poisson-Gaussian noise model. We also use the properties of the variance stabilizer to develop a principle for selection of the regularization parameter in penalized estimators for Poisson and Poisson-Gaussian inverse problems.

MSC:

94A12 Signal theory (characterization, reconstruction, filtering, etc.)
62H12 Estimation in multivariate analysis
62H35 Image analysis in multivariate analysis

Software:

SPIRAL; CVX
PDFBibTeX XMLCite
Full Text: DOI Link

References:

[1] Candes E 2008 The restricted isometry property and its implications for compressed sensing C. R. Math.346 589-92 · Zbl 1153.94002 · doi:10.1016/j.crma.2008.03.014
[2] Trussell H J and Zhang R 2012 The dominance of Poisson noise in color digital cameras ICIP pp 329-32
[3] Murtagh F, Starck J-L and Bijaoui A 1995 Image restoration with noise suppression using a multiresolution support Astron. Astrophys.112 179
[4] Delpretti S, Luisier F, Ramani S, Blu T and Unser M 2008 Multiframe sure-let denoising of timelapse fluorescence microscopy images ISBI pp 149-52
[5] Du Bosq T and Preece B 2016 Performance assessment of a singlepixel compressive sensing imaging system Proc. SPIE9820 98200F
[6] Duarte M, Davenport M, Takhar D, Laska J, Sun T, Kelly K and Baraniuk R 2008 Single pixel imaging via compressive sampling IEEE Signal Process. Mag.25 83-92 · doi:10.1109/MSP.2007.914730
[7] Chouzenoux E, Jezierska A, Pesquet J-C and Talbot H 2015 A convex approach for image restoration with exact Poisson-Gaussian likelihood SIAM J. Imaging Sci.8 2662-82 · Zbl 1343.94004 · doi:10.1137/15M1014395
[8] Luisier F, Blu T and Unser M 2011 Image denoising in mixed Poisson-Gaussian noise IEEE TIP20 696-708 · Zbl 1372.94168
[9] Shin D, Shapiro J-H and Goyal V 2016 Performance analysis of low-flux least-squares single-pixel imaging IEEE Signal Process. Lett.23 1756-60 · doi:10.1109/LSP.2016.2617329
[10] Harmany Z et al 2012 This is SPIRAL-TAP: sparse Poisson intensity reconstruction algorithms—theory and practice IEEE TIP21 1084-96 · Zbl 1372.94381
[11] Lingenfelter D, Fessler J and He Z 2009 Sparsity regularization for image reconstruction with Poisson data Proc. SPIE7246
[12] Starck J L and Bobin J 2010 Astronomical data analysis and sparsity: from wavelets to compressed sensing Proc. IEEE98 1021-30 · doi:10.1109/JPROC.2009.2025663
[13] Zhang B, Fadili M and Starck J 2008 Wavelets, ridgelets, and curvelets for Poisson noise removal IEEE TIP17 1093-108
[14] Dupé F-X, Fadili M and Starck J 2009 A proximal iteration for deconvolving Poisson noisy images using sparse representations IEEE Trans. Image Process.18 310-21 · Zbl 1371.94117 · doi:10.1109/TIP.2008.2008223
[15] Jeong T, Woo H and Yun S 2013 Frame-based poisson image restoration using a proximal linearized alternating direction method Inverse Problems29 075007 · Zbl 1322.94017 · doi:10.1088/0266-5611/29/7/075007
[16] Hohage T and Werner F 2016 Inverse problems with Poisson data: statistical regularization theory, applications and algorithms Inverse Problems32 093001 · Zbl 1372.65163 · doi:10.1088/0266-5611/32/9/093001
[17] Anscombe F J 1948 The transformation of Poisson, binomial and negative-binomial data Biometrika35 246-54 · Zbl 0032.03702 · doi:10.2307/2332343
[18] Garg D and Rajwade A 2017 Performance bounds for Poisson compressed sensing using variance-stabilization transforms ICASSP pp 1-4
[19] Raginsky M, Willett R, Harmany Z and Marcia R 2010 Compressed sensing performance bounds under Poisson noise IEEE TSP58 3990-4002 · Zbl 1392.94417
[20] Baraniuk R, Davenport M, DeVore R and Wakin M 2008 A simple proof of the restricted isometry property for random matrices Constructive Approx.28 253-63 · Zbl 1177.15015 · doi:10.1007/s00365-007-9003-x
[21] Bobin J, Starck J-L and Ottensamer R 2008 Compressed sensing in astronomy IEEE J. Sel. Top. Signal Process.2 718-26 · doi:10.1109/JSTSP.2008.2005337
[22] Scaife A, Puy G, Jacques L, Vandergheynst P and Wiaux Y 2009 Compressed sensing imaging techniques for radio interferometry Mon. Not. R. Astron. Soc.395 1733-42 · doi:10.1111/j.1365-2966.2009.14665.x
[23] Studer V, Bobin J, Chahid M, Mousavi H-S, Candes E and Dahan M 2012 Compressive fluorescence microscopy for biological and hyperspectral imaging Proc. Natl Acad. Sci.109 · doi:10.1073/pnas.1119511109
[24] Howland G, Lum D, Ware M and Howell J 2013 Photon counting compressive depth mapping Opt. Express21 23822-37 · doi:10.1364/OE.21.023822
[25] Ma J 2009 Single-pixel remote sensing IEEE Geosci. Remote Sens. Lett.6 199-203 · doi:10.1109/LGRS.2008.2010959
[26] Curtiss J H 1943 On transformations used in the analysis of variance Ann. Math. Stat.14 107-22 · Zbl 0060.31203 · doi:10.1214/aoms/1177731452
[27] Code for reproducing results in this paper (https://www.cse.iitb.ac.in/ ajitvr/Poisson_PoissonGaussian_CS/)
[28] Bertero M, Boccacci P, Talenti G, Zanella R and Zanni L 2010 A discrepancy principle for Poisson data Inverse Problems26 105004 · Zbl 1200.62067 · doi:10.1088/0266-5611/26/10/105004
[29] Jiang X, Raskutti G and Willett R 2015 Minimax optimal rates for Poisson inverse problems with physical constraints IEEE TIT61 4458-74 · Zbl 1359.94106
[30] Bartlett M S 1936 The square root transformation in the analysis of variance J. R. Stat. Soc.68 1 · JFM 63.1085.01
[31] Candes E, Romberg J and Tao T 2006 Stable signal recovery from incomplete and inaccurate measurements Commun. Pure Appl. Math.59 1207-23 · Zbl 1098.94009 · doi:10.1002/cpa.20124
[32] Candes E and Tao T 2007 The dantzig selector: statistical estimation when p is much larger than n Ann. Stat.35 2313-51 · Zbl 1139.62019 · doi:10.1214/009053606000001523
[33] Freeman M and Tukey J 1950 Transformations related to the angular and the square root Ann. Math. Stat.21 607-11 · Zbl 0039.35304 · doi:10.1214/aoms/1177729756
[34] Laurent B and Massart P 2000 Adaptive estimation of a quadratic functional by model selection Ann. Stat.28 1302-38 · Zbl 1105.62328 · doi:10.1214/aos/1015957395
[35] Boyd S and Vandenberghe L 2004 Convex Optimization (Cambridge: Cambridge University Press) · Zbl 1058.90049 · doi:10.1017/CBO9780511804441
[36] Anzengruber S and Ramlau R 2009 Morozovs discrepancy principle for tikhonov-type functionals with nonlinear operators Inverse Problems26 025001 · Zbl 1192.65064 · doi:10.1088/0266-5611/26/2/025001
[37] Heath M 1996 Scientific Computing: an Introductory Survey 2nd edn (New York: McGraw-Hill) · Zbl 0903.68072
[38] Zanni L, Benfenati A, Bertero M and Ruggiero V 2015 Numerical methods for parameter estimation in poisson data inversion J. Math. Imaging Vis.52 397-413 · Zbl 1327.65117 · doi:10.1007/s10851-014-0553-9
[39] Grant M and Boyd S 2014 CVX: Matlab software for disciplined convex programming version 2.1 (http://cvxr.com/cvx)
[40] Rohban M-H, Saligrama V and Vaziri D-M 2016 Minimax optimal sparse signal recovery with Poisson statistics IEEE TSP64 3495-508 · Zbl 1414.94511
[41] Li Y and Raskutti G 2018 Minimax optimal convex methods for Poisson inverse problems under lq-ball sparsity IEEE Transactions on Information Theory64 8 · Zbl 1401.90260 · doi:10.1109/TIT.2018.2850365
[42] Frick K, Marnitz P and Munk A 2013 Statistical multiresolution estimation for variational imaging: with an application in Poisson-biophotonics J. Math. Imaging Vis.46 370-87 · Zbl 1312.68222 · doi:10.1007/s10851-012-0368-5
[43] Li Y-H and Cevher V 2015 Consistency of l1-regularized maximum-likelihood for compressive Poisson regression ICASSP pp 3606-10
[44] Bardsley J and Goldes J 2009 Regularization parameter selection methods for ill-posed poisson maximum likelihood estimation Inverse Problems25 095005 · Zbl 1176.68225 · doi:10.1088/0266-5611/25/9/095005
[45] Oike Y and El Gamal A 2013 CMOS image sensor with per-column sigma delta ADC and programmable compressed sensing IEEE J. Solid-State Circuits48 318-28 · doi:10.1109/JSSC.2012.2214851
[46] Kulkarni K, Lohit S, Turaga P, Kerviche R and Ashok A 2016 Reconnet: non-iterative reconstruction of images from compressively sensed measurements CVPR
[47] Laska J, Boufounos P, Davenport M and Baraniuk R 2011 Democracy in action: quantization, saturation, and compressive sensing Appl. Comput. Harmon. Anal.31 429-43 · Zbl 1231.94045 · doi:10.1016/j.acha.2011.02.002
[48] Pollard J H 1977 A Handbook of Numerical and Statistical Techniques: with Examples Mainly from the Life Sciences (Cambridge: Cambridge University Press) · Zbl 0357.62090 · doi:10.1017/CBO9780511569692
[49] Patil S and Rajwade A 2016 Poisson noise removal for image demosaicing Proc. of the British Machine Vision Conf. · doi:10.5244/C.30.33
[50] Jiang X, Reynaud-Bouret P, Rivoirard V, Sansonnet L and Willett R 2016 A data-dependent weighted LASSO under Poisson noise 65 1589-613 · Zbl 1432.62222 · doi:10.1109/TIT.2018.2869578
[51] Patil S, Gurumoorthy K and Rajwade A 2019 Using an information theoretic metric for compressive recovery under poisson noise Signal Process.162 35-53 · doi:10.1016/j.sigpro.2019.04.004
[52] Rish I and Grabarnik G 2009 Sparse signal recovery with exponential-family noise ACCS pp 60-6
[53] Kakade S, Shamir O, Sindharan K and Tewari A 2010 Learning exponential families in high-dimensions: Strong convexity and sparsity AISTATS pp 381-8
[54] Jinzhu J, Karl R and Bin Y 2013 The LASSO under Poisson-like heterscedasticity Stat. Sin.23 99-118 · Zbl 1259.62042
[55] Ivanoff S, Picard F and Rivoirard V 2016 Adaptive LASSO and group-LASSO for functional Poisson regression JMLR17 1-46 · Zbl 1360.62398
[56] Foucart S and Rauhut H 2013 A Mathematical Introduction to Compressive Sensing (Basel: Birkhäuser) · Zbl 1315.94002 · doi:10.1007/978-0-8176-4948-7
[57] Plan Y and Vershynin R 2016 The generalized LASSO with non-linear observations IEEE TIT62 1528-37 · Zbl 1359.94153
[58] Yang Z, Wang Z, Liu H, Eldar Y and Zhang T 2015 Sparse nonlinear regression: Parameter estimation and asymptotic inference (arXiv:1511.04514)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.