Compressive Deconvolution in Random Mask Imaging
- 1 October 2015
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 1 (4) , 236-246
Abstract
We investigate the problem of reconstructing signals from a subsampled convolution of their modulated versions and a known filter. The problem is studied as applies to a specific imaging architecture that relies on spatial phase modulation by randomly coded “masks.” The diversity induced by the random masks is deemed to improve the conditioning of the deconvolution problem while maintaining sampling efficiency. We analyze a linear model of the imaging system, where the joint effect of the spatial modulation, blurring, and spatial subsampling is represented concisely by a measurement matrix. We provide a bound on the conditioning of this measurement matrix in terms of the number of masks K, the dimension (i.e., the pixel count) of the scene image L, and certain characteristics of the blurring kernel and subsampling operator. The derived bound shows that the stable deconvolution is possible with high probability even if the number of masks (i.e., K) is as small as L log L /N , meaning that the total number of (scalar) measurements is within a logarithmic factor of the image size. Furthermore, beyond a critical number of masks determined by the extent of blurring and subsampling, use of every additional mask improves the conditioning of the measurement matrix. We also consider a more interesting scenario where the target image is known to be sparse. We show that under mild conditions on the blurring kernel, with high probability the measurement matrix is a restricted isometry when the number of masks is within a logarithmic factor of the sparsity of the scene image. Therefore, the scene image can be reconstructed using any of the well-known sparse recovery algorithms such as the basis pursuit. The bound on the required number of masks grows linearly in sparsity of the scene image but logarithmically in its ambient dimension. The bound provides a quantitative view of the effect of the blurring and subsampling on the required number of masks, which is critical for designing efficient imaging systems.Keywords
All Related Versions
Funding Information
- ONR (N00014-11-1-0459)
- NSF (CCF-1415498, CCF-1422540)
This publication has 20 references indexed in Scilit:
- 3‐D PSF fitting for fluorescence microscopy: implementation and localization applicationJournal of Microscopy, 2012
- Restricted isometries for partial random circulant matricesApplied and Computational Harmonic Analysis, 2012
- Sparse Recovery Algorithms: Sufficient Conditions in Terms of Restricted Isometry Constants2011
- User-Friendly Tail Bounds for Sums of Random MatricesFoundations of Computational Mathematics, 2011
- Compressed sensing for practical optical imaging systems: A tutorial2011
- A note on guaranteed sparse recovery via -minimizationApplied and Computational Harmonic Analysis, 2010
- Compressive Sensing by Random ConvolutionSIAM Journal on Imaging Sciences, 2009
- The restricted isometry property and its implications for compressed sensingComptes Rendus Mathématique, 2008
- Single-pixel imaging via compressive samplingIEEE Signal Processing Magazine, 2008
- Digital image restorationIEEE Signal Processing Magazine, 1997