Abstract
An interesting problem arising in astronomical imaging is the reconstruction of an image with high dynamic range, for example a set of bright point sources superimposed to smooth structures. A few methods have been proposed for dealing with this problem and their performance is not always satisfactory. In this paper we propose a solution based on the representation, already proposed elsewhere, of the image as the sum of a pointwise component and a smooth one, with different regularization for the two components. Our approach is in the framework of Poisson data and to this purpose we need efficient deconvolution methods. Therefore, first we briefly describe the application of the Scaled Gradient Projection (SGP) method to the case of different regularization schemes and subsequently we propose how to apply these methods to the case of multiple image deconvolution of high-dynamic range images, with specific reference to the Fizeau interferometer LBTI of the Large Binocular Telescope (LBT). The efficacy of the proposed methods is illustrated both on simulated images and on real images, observed with LBTI, of the Jovian moon Io. The software is available at http://www.oasis.unimore.it/site/home/software.html.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
AIRY can be downloaded from http://www.airyproject.eu.
- 2.
- 3.
The LBT is an international collaboration among institutions in the United States, Italy and Germany. LBT Corporation partners are: The University of Arizona on behalf of the Arizona Board of Regents; Istituto Nazionale di Astrofisica, Italy; LBT Beteiligungsgesellschaft, Germany, representing the Max-Planck Society, the Leibniz Institute for Astrophysics Potsdam, and Heidelberg University; the Ohio State University, and the Research Corporation, on behalf of the University of Notre Dame, University of Minnesota and University of Virginia.
References
Anconelli, B., Bertero, M., Boccacci, P., Carbillet, M., Lantéri, H.: Reduction of boundary effects in multiple image deconvolution with an application to LBT LINC-NIRVANA. Astron. Astrophys. 448, 1217–1224 (2006)
Bardsley, J.M., Goldes, J.: Regularization parameter selection methods for ill-posed poisson maximum likelihood estimation. Inverse Probl. 25, 095,005 (2009)
Barzilai, J., Borwein, J.M.: Two-point step size gradient methods. IMA J. Numer. Anal. 8(1), 141–148 (1988)
Benfenati, A., Ruggiero, V.: Inexact bregman iteration with an application to poisson data reconstruction. Inverse Probl. 29, 065,016 (2013)
Bertero, M., Boccacci, P.: Introduction to Inverse Problems in Imaging. IoP Publishing, Bristol (1998)
Bertero, M., Boccacci, P.: A simple method for the reduction of boundary effects in the Richardson-Lucy approach to image deconvolution. Astron. Astrophys. 437, 369–374 (2005)
Bertero, M., Boccacci, P., Desiderà, G., Vicidomini, G.: Image deblurring with Poisson data: from cells to galaxies. Inverse Probl. 25(12), 123,006 (2009)
Bertero, M., Boccacci, P., La Camera, A., Olivieri, C., Carbillet, M.: Imaging with LINC-NIRVANA, the Fizeau interferometer of the Large Binocular Telescope: state of the art and open problems. Inverse Probl. 27(11), 113,011 (2011)
Bertero, M., Boccacci, P., Talenti, G., Zanella, R., Zanni, L.: A discrepancy principle for Poisson data. Inverse Probl. 26(10), 10,500 (2010)
Bertsekas, D.: Nonlinear Programming. Athena Scientific, Belmont (1999)
Bonettini, S., Landi, G., Loli Piccolomini, E., Zanni, L.: Scaling techniques for gradient projection-type methods in astronomical image deblurring. Int. J. Comput. Math. 90(1), 9–29 (2013)
Bonettini, S., Prato, M.: Nonnegative image reconstruction from sparse Fourier data: a new deconvolution algorithm. Inverse Probl. 26(9), 095,001 (2010)
Bonettini, S., Prato, M.: New convergence results for the scaled gradient projection method. Inverse Probl. 31(9), 095,008 (2015)
Bonettini, S., Ruggiero, V.: On the uniqueness of the solution of image reconstruction problems with Poisson data. In: Simos, T.E., Psihoyios, G., Tsitouras, C. (eds.), International Conference of Numerical Analysis and Applied Mathematics 2010, AIP Conference Proceedings, vol. 1281, pp. 1803–1806 (2010)
Bonettini, S., Ruggiero, V.: An alternating extragradient method for total variation-based image restoration. Inverse Problems 27, 095,001 (2011)
Bonettini, S., Zanella, R., Zanni, L.: A scaled gradient projection method for constrained image deblurring. Inverse Probl. 25(1), 015,002 (2009)
Byrne, C.L.: Iterative image reconstruction algorithms based on cross-entropy minimization. IEEE Trans. Image Proc. 2(1), 96–103 (1993)
Carasso, A.S.: Linear and nonlinear image deblurring: a documented study. SIAM J. Numer. Anal. 36(6), 1659–1689 (1999)
Carbillet, M., La Camera, A., Deguignet, J., Prato, M., Bertero, M., Aristidi, E., Boccacci, P.: Strehl-constrained reconstruction of post-adaptive optics data and the Software Package AIRY, v. 6.1. In: Marchetti, E., Close, L.M., Véran, J.P. (eds.), Adaptive Optics Systems IV. Proceedings of SPIE, vol. 9148, p. 91484U (2014)
Charbonnier, P., Blanc-Féraud, L., Aubert, G., Barlaud, M.: Deterministic edge-preserving regularization in computed imaging. IEEE T. Image Process. 6, 298–311 (1997)
Conrad, A., de Kleer, K., Leisenring, J., La Camera, A., Arcidiacono, C., Bertero, M., Boccacci, P., Defrère, D., de Pater, I., Hinz, P., Hofmann, K.H., Kürster, M., Rathbun, J., Schertl, D., Skemer, A., Skrutskie, M., Spencer, J., Veillet, C., Weigelt, G., Woodward, C.E.: Spatially Resolved M-Band Emission from Io’s Loki Patera-Fizeau Imaging at the 22.8m LBT. Astron. J. 149(5), 1–9 (2015)
Cornelio, A., Porta, F., Prato, M.: A convergent least-squares regularized blind deconvolution approach. Appl. Math. Comput. 259(12), 173–186 (2015)
Correia, S., Carbillet, M., Boccacci, P., Bertero, M., Fini, L.: Restoration of interferometric images: I. the software package AIRY. Astron. Astrophys. 387, 733–743 (2002)
Csiszár, I.: Why least squares and maximum entropy? An axiomatic approach to inference for linear inverse problems. Ann. Stat. 19(4), 2032–2066 (1991)
De Mol, C., Defrise, M.: Inverse imaging with mixed penalties. In: Proceedings of the International Symposium on Electromagnetic Theory, pp. 798–800. Pisa, Italy (2004)
Engl, H.W., Hanke, M., Neubauer, A.: Regularization of Inverse Problems. Kluver Academic Publishers, Dordrecht (1996)
Geman, S., Geman, D.: Stochastic relaxation, Gibbs distributions and the Bayesian restoration of images. IEEE Trans. Pattern Anal. Mach. Intell. 6(6), 721–741 (1984)
Giovannelli, J.F., Coulais, A.: Positive deconvolution for superimposed extended source and point sources. Astron. Astrophys. 439, 401–412 (2005)
Herbst, T., Ragazzoni, R., Andersen, D., Boehnhardt, H., Bizenberger, P., Eckart, A., Gaessler, W., Rix, H.W., Rohloff, R.R., Salinari, P., Soci, R., Straubmeier, C., Xu, W.: LINC-NIRVANA: a fizeau beam combiner for the large binocular telescope. In: W.A. Traub (ed.), Interferometry for Optical Astronomy II. Proceedings of SPIE, vol. 4838, pp. 456–465 (2003)
Herbst, T.M., Santhakumari, K.K.R., Klettke, M., Arcidiacono, C., Bergomi, M., Bertram, T., Berwein, J., Bizenberger, P., Briegel, F., Farinato, J., Marafatto, L., Mathar, R., McGurk, R., Ragazzoni, R., Viotto, V.: Commissioning multi-conjugate adaptive optics with LINC-NIRVANA on LBT. In: Adaptive Optics Systems VI. Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, vol. 10703, p. 107030B (2018). https://doi.org/10.1117/12.2313421
Hill, J.M., Green, R.F., Slagle, J.H.: The large binocular telescope. In: Ground-based and Airborne Telescopes. Proceedings of SPIE, vol. 6267, p. 62670Y (2006)
Hinz, P., Bippert-Plymate, T., Breuninger, A., Connors, T., Duffy, B., Esposito, S., Hoffmann, W., Kim, J., Kraus, J., McMahon, T., Montoya, M., Nash, R., Durney, O., Solheid, E., Tozzi, A., Vaitheeswaran, V.: Status of the LBT interferometer. In: Optical and Infrared Interferometry. Proceedings of SPIE, vol. 7013, p. 701328 (2008)
Hom, E.F.Y., Marchis, F., Lee, T.K., Haase, S., Agard, D.A., Sedat, J.W.: Aida: an adaptive image deconvolution algorithm with application to multi-frame and three-dimensional data. J. Opt. Soc. Am. A-24, 1580–1600 (1994)
Jefferies, S.M., Christou, J.C.: Restoration of astronomical images by iterative blind deconvolution. Astron. J. 415, 609–862 (1993)
La Camera, A., Antonucci, S., Bertero, M., Boccacci, P., Lorenzetti, D., Nisini, B.: Image reconstruction for observations with a high dynamic range: LINC-NIRVANA simulations of a stellar jet. In: Delplancke, F., Rajagopal, F.J.K., Malbet, F. (eds.) Optical and Infrared Interferometry III. Proceedings of SPIE, vol. 8455, p. 84553D (2012)
Lantéri, H., Roche, M., Aime, C.: Penalized maximum likelihood image restoration with positivity constraints: multiplicative algorithms. Inverse Probl. 18(5), 1397–1419 (2002)
Leisenring, J.M., Hinz, P.M., Skrutskie, M.F., Skemer, A., Woodward, C.E., Veillet, C., Arcidiacono, C., Bailey, V., Bertero, M., Boccacci, P., Conrad, A., de Kleer, K., de Pater, I., Defrère, D., Hill, J., Hofmann, K.H., Kaltenegger, L., La Camera, A., Nelson, M.J., Schertl, D., Spencer, J., Weigelt, G., Wilson, J.C.: Fizeau interferometric imaging of Io volcanism with LBTI/LMIRcam. In: Optical and Infrared Interferometry IV. Proceedings of SPIE, vol. 9146, p. 91462S (2014)
Mugnier, L.M., Fusco, T., Conan, J.M.: Mistral: a myopic edge-preserving image restoration method, with application to astronomical adaptive-optics-corrected long-exposure images. J. Opt. Soc. Am. A-21(4), 1841–1854 (2004)
Porta, F., Prato, M., Zanni, L.: A new steplength selection for scaled gradient methods with application to image deblurring. J. Sci. Comput. 65(3), 895–919 (2015)
Porta, F., Zanella, R., Zanghirati, G., Zanni, L.: Limited-memory scaled gradient projection methods for real-time image deconvolution in microscopy. Commun. Nonlinear Sci. Numer. Simul. 21(1–3), 112–127 (2015)
Prato, M., Cavicchioli, R., Zanni, L., Boccacci, P., Bertero, M.: Efficient deconvolution methods for astronomical imaging: algorithms and IDL-GPU codes. Astron. Astrophys. 539, A133 (2012)
Prato, M., La Camera, A., Bonettini, S., Bertero, M.: A convergent blind deconvolution method for post-adaptive-optics astronomical imaging. Inverse Probl. 29(6), 065,017 (2013)
Prato, M., La Camera, A., Bonettini, S., Rebegoldi, S., Bertero, M., Boccacci, P.: A blind deconvolution method for ground based telescopes and fizeau interferometers. New Astron. 40, 1–13 (2015)
Skilling, J., Bryan, R.K.: Maximum entropy image reconstruction: general algorithm. Mon. Not. R. Astr. Soc. 211, 111–124 (1984)
Skrutskie, M.F., Jones, T., Hinz, P., Garnavich, P., Wilson, J., Nelson, M., Solheid, E., Durney, O., Hoffmann, W.F., Vaitheeswaran, V., McMahon, T., Leisenring, J., Wong, A., Garnavic, P., Wilson, J., Nelson, M., Sollheid, E., Durney, O., Hoffmann, W.F., Vaitheeswaran, V., McMahon, T., Leisenring, J., Wong, A.: The Large Binocular Telescope mid-infrared camera (LMIRcam): final design and status. In: McLean, I.S., Ramsay, S.K., Takami, H. (eds.) Ground-based and Airborne Instrumentation for Astronomy III. Proceedings of SPIE, vol. 7735, p. 77353H (2010). https://doi.org/10.1117/12.857615. http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=750817
Snyder, D.L., Hammoud, A.M., White, R.L.: Image recovery from data acquired with a charge-coupled-device camera. J. Opt. Soc. Am. A 10(5), 1014–1023 (1994)
Snyder, D.L., Helstrom, C.W., Lanterman, A.D., Faisal, M., White, R.L.: Compensation for readout noise in CCD images. J. Opt. Soc. Am. A 12(2), 272–283 (1995)
Titterington, D.: On the iterative image space reconstruction algorithm for ECT. IEEE Trans. Med. Imaging 6(1), 52–56 (1987)
Vio, R., Bardsley, J., Wamsteher, W.: Least-squares methods with poissonian noise: analysis and comparison with the richardson-lucy algorithm. Astron. Astrophys. 436(2), 741–755 (2005)
Zanella, R., Boccacci, P., Zanni, L., Bertero, M.: Efficient gradient projection methods for edge-preserving removal of Poisson noise. Inverse Probl. 25(4), 045,010 (2009)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
Appendix
Regularization functions
In this paper we assume that \({\varvec{f}}\) is a \(N \times N\) array extended (when needed) with periodic boundary conditions, i.e., if we set \({\varvec{n}} = (n_1, n_2)\), then \({\varvec{f}}(N+1,n_2) = {\varvec{f}}(1,n_2)\), \({\varvec{f}}(n_1,N+1) = {\varvec{f}}(n_1,1)\) and \({\varvec{f}}(N+1,N+1) = {\varvec{f}}(1,1)\).
For introducing the regularization functions considered in our methods and software we need some notation. We set \({\varvec{n}}_{1 \pm } = (n_1 \pm 1, n_2)\) and \({\varvec{n}}_{2 \pm } = (n_1, n_2 \pm 1)\) and we introduce the square and the modulus of the discrete gradient
Then, the seven regularization functions and the corresponding arrays \(U_1,V_1\) are the following:
-
Zeroth order Tikhonov (T-0) regularization
$$\begin{aligned} J_1({\varvec{f}}) = \frac{1}{2} \sum _{\varvec{n}} |{\varvec{f}}({\varvec{n}})|^2~~, \end{aligned}$$(29)for which (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) = 0~~,~~{\varvec{V}}_1({\varvec{n}}) = {\varvec{f}}({\varvec{n}}). \end{aligned}$$(30) -
First order Tikhonov (T-1) regularization
$$\begin{aligned} J_1({\varvec{f}}) = \frac{1}{2} \sum _{{\varvec{n}}} {\varvec{D}}^2({\varvec{n}})~~, \end{aligned}$$(31)for which (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) =&~{\varvec{f}}({\varvec{n}}_{1+})+{\varvec{f}}({\varvec{n}}_{2+})+{\varvec{f}}({\varvec{n}}_{1-})+{\varvec{f}}({\varvec{n}}_{2-})~~, \\ \nonumber {\varvec{V}}_1({\varvec{n}}) =&~4 {\varvec{f}}({\varvec{n}})~~. \end{aligned}$$ -
Second order Tikhonov (T-2) regularization
$$\begin{aligned} J_1({\varvec{f}}) = \frac{1}{2} \sum _{\varvec{n}}(\varDelta {\varvec{f}})({\varvec{n}})^2~~, \end{aligned}$$(32)where \(\varDelta \) denotes the discrete Laplacian. As remarked in [36], it can be written in the form
$$\begin{aligned} J_1({\varvec{f}}) = \frac{1}{2} \sum _{\varvec{n}} \big [{\varvec{f}}({\varvec{n}}) -(B{\varvec{f}})({\varvec{n}})\big ]^2~~, \end{aligned}$$(33)where B is the convolution matrix obtained from the \(3 \times 3\) mask with columns (0, 1/4, 0), (1/4, 0, 1/4) and (0, 1/4, 0). Then (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) =&~[(B + B^T) {\varvec{f}}]({\varvec{n}})~~, \\ \nonumber {\varvec{V}}_1({\varvec{n}}) =&~[(I + B^TB){\varvec{f}}]({\varvec{n}})~~. \end{aligned}$$ -
Cross-Entropy (CE) regularization [17, 18]
$$\begin{aligned} J_1({\varvec{f}}) = KL({\varvec{f}}, \bar{\varvec{f}}) = \sum _{{\varvec{n}}} \big \{{\varvec{f}}({\varvec{n}})\ln \left( \frac{{\varvec{f}}({\varvec{n}})}{\bar{\varvec{f}}({\varvec{n}})}\right) + \bar{\varvec{f}}({\varvec{n}}) - {\varvec{f}}({\varvec{n}})\big \}~~, \end{aligned}$$(34)where \(\bar{\varvec{f}}\) is a reference image. When \(\bar{\varvec{f}}\) is a constant array, then the cross-entropy becomes the negative Shannon entropy considered, for instance, in [44]. If both \({\varvec{f}}\) and \(\bar{\varvec{f}}\) satisfy the constraint (4), then a possible choice for the functions \(U_1,~V_1\) is
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) = - \ln \frac{{\varvec{f}}({\varvec{n}})}{c}~~,~~{\varvec{V}}_1({\varvec{n}}) = - \ln \frac{\bar{\varvec{f}}({\varvec{n}})}{c}~~, \end{aligned}$$(35)where c is the flux constant defined in (4). We remark that, since the background is taken into account by the algorithms, \({\varvec{f}}\) can be zero in some pixels; for this reason in the computation of the gradient we add a small quantity to the values of \({\varvec{f}}\). We also remark that when \(\bar{\varvec{f}}\) is a constant, e.g. \(c/N^2\), then \({\varvec{V}}_1({\varvec{n}}) = 2 \mathrm{ln}~N\).
-
Hypersurface (HS) regularization [20]
$$\begin{aligned} J_1({\varvec{f}}) = \sum _{{\varvec{n}}} \sqrt{\delta ^2 + {\varvec{D}}^2({\varvec{n}})}~~,~~\delta > 0~~, \end{aligned}$$(36)for which (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) = ~\frac{{\varvec{f}}({\varvec{n}}_{1+}) \!+\! {\varvec{f}}({\varvec{n}}_{2+})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}})}} \!+\! \frac{{\varvec{f}}({\varvec{n}}_{1-})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}}_{1-})}} \!+\! \frac{{\varvec{f}}({\varvec{n}}_{2-})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}}_{2-})}}~,\\ \nonumber {\varvec{V}}_1({\varvec{n}}) = ~\frac{2 {\varvec{f}}({\varvec{n}})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}})}} + \frac{{\varvec{f}}({\varvec{n}})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}}_{1-})}} + \frac{{\varvec{f}}({\varvec{n}})}{\sqrt{\delta ^2 \!+\! {\varvec{D}}^2({\varvec{n}}_{2-})}}~. \end{aligned}$$The application of SGP to the case of HS regularization is already considered in [15] and [4] for a comparison of its accuracy with that of Total Variation (TV) regularization.
-
Markov random field (MRF) regularization [27]
$$\begin{aligned} J_1({\varvec{f}}) = \frac{1}{2}\sum _{{\varvec{n}}} \sum _{{\varvec{n}}' \in \mathcal{N}({\varvec{n}})}\sqrt{\delta ^2 + \left( \frac{{\varvec{f}}({\varvec{n}})-{\varvec{f}}({\varvec{n}}')}{\epsilon ({\varvec{n}}')}\right) ^2}~~, \end{aligned}$$(37)where \(\delta >0\), \(\mathcal{N}({\varvec{n}})\) is a symmetric neighborhood made up of the eight first neighbors of \({\varvec{n}}\) and \(\epsilon ({\varvec{n}}')\) is equal to 1 for the horizontal and vertical neighbors and equal to \(\sqrt{2}\) for the diagonal ones; thanks to the symmetry of \(\mathcal{N}({\varvec{n}})\), Eq. (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) =&~\sum _{{\varvec{n}}' \in \mathcal{N}({\varvec{n}})}\frac{{\varvec{f}}({\varvec{n}}')}{\epsilon ({\varvec{n}}')\sqrt{\delta ^2 + \left( \frac{{\varvec{f}}({\varvec{n}}')-{\varvec{f}}({\varvec{n}}')}{\epsilon ({\varvec{n}}')}\right) ^2}}~~, \\ \nonumber {\varvec{V}}_1({\varvec{n}}) =&~\sum _{{\varvec{n}}' \in \mathcal{N}({\varvec{n}})}\frac{{\varvec{f}}({\varvec{n}})}{\epsilon ({\varvec{n}}')\sqrt{\delta ^2 + \left( \frac{{\varvec{f}}({\varvec{n}})-{\varvec{f}}({\varvec{n}}')}{\epsilon ({\varvec{n}}')}\right) ^2}}~~. \end{aligned}$$ -
MISTRAL regularization (MIST) [38]
$$\begin{aligned} J_1({\varvec{f}}) = \sum _{{\varvec{n}}} \left\{ |{\varvec{D}}({\varvec{n}})| -\delta \ln \left( 1 + \frac{|{\varvec{D}}({\varvec{n}})|}{\delta }\right) \right\} ~~,~~\delta > 0~~, \end{aligned}$$(38)for which (5) holds by setting
$$\begin{aligned} {\varvec{U}}_1({\varvec{n}}) =&~\frac{{\varvec{f}}({\varvec{n}}_{1+}) + {\varvec{f}}({\varvec{n}}_{2+})}{\delta + |{\varvec{D}}({\varvec{n}})|} +\frac{{\varvec{f}}({\varvec{n}}_{1-})}{\delta + |{\varvec{D}}({\varvec{n}}_{1-})|} + \frac{{\varvec{f}}({\varvec{n}}_{2-})}{\delta + |{\varvec{D}}({\varvec{n}}_{2-})|}~~, \\ \nonumber {\varvec{V}}_1({\varvec{n}}) =&~\frac{2{\varvec{f}}({\varvec{n}})}{\delta + |{\varvec{D}}({\varvec{n}})|} + \frac{{\varvec{f}}({\varvec{n}})}{\delta + |{\varvec{D}}({\varvec{n}}_{1-})|} + \frac{{\varvec{f}}({\varvec{n}})}{\delta + |{\varvec{D}}({\varvec{n}}_{2-})|}~~. \end{aligned}$$
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Prato, M., La Camera, A., Arcidiacono, C., Boccacci, P., Bertero, M. (2019). Multiple Image Deblurring with High Dynamic-Range Poisson Data. In: Donatelli, M., Serra-Capizzano, S. (eds) Computational Methods for Inverse Problems in Imaging. Springer INdAM Series, vol 36. Springer, Cham. https://doi.org/10.1007/978-3-030-32882-5_6
Download citation
DOI: https://doi.org/10.1007/978-3-030-32882-5_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-32881-8
Online ISBN: 978-3-030-32882-5
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)