Variance-stabilization-based compressive inversion under Poisson or Poisson–Gaussian noise with analytical bounds

Abstract

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 nonnegativity 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.

A Compressed Sensing Approach to Pooled RT-PCR Testing for COVID-19 Detection

“…4, the expected value of R(y, Ax) := log y − log Ax 2 should be close to σ log(1 + q) √ m. Moreover, its variance is quite small and independent of m. Hence, as per the discrepancy principle, we seek to find λ ∈ Λ such that |R(y, A xλ )−σ log(1+q) √ m| is minimized. Previous work in [18] has employed such a technique in the context of image deblurring under Poisson-Gaussian noise with a square-root based data-fidelity term of the form y + 3/8 − Ax λ + 3/8 2 .…”

A novel variational approach for multiphoton microscopy image restoration: from PSF estimation to 3D deconvolution

“…We fitted a convolution kernel on 8 cropped volumes, containing isolated bead images, using GENTLE (algorithm 1). The hyper-parameters of the PSF estimation model (6) were set identically to section 2.5. Figure 7 presents a comparison of the estimated FWHMs along the principal axis and Euler angles for the 8 beads, using the GENTLE method.…”

“…However, such a noise model can be tedious to deal with numerically at large-scale, often involving minimizing a non-smooth data-fidelity term [16,79]. While this issue has been tackled in confocal microscopy [6,72,79], its exploration in MPM imaging remains scarce.…”

Reconstruction Of Sparse Signals Using Likelihood Maximization from Compressive Measurements with Gaussian And Saturation Noise

“…We have proved the convexity of our estimator and derived the upper performance bound, and shown that it numerically outperforms competing methods. The recent work in [3] handles compressive inversion under with Poisson-Gaussian-uniform quantization noise, a very realistic noise model in imaging systems. Extending the numerical simulations as well as the convexity proofs to handle saturation effects in conjunction with such a Poisson-Gaussian noise model is a potential avenue for future work.…”