Abstract
We leverage convex optimization techniques to perform
Bayesian uncertainty quantification (UQ) for hyperspectral (HS) inverse
imaging problems. The proposed approach generalizes our recent work
for single-channel UQ [1]. Similarly, the Bayesian hypothesis test is
formulated as a convex minimization problem and solved using a primaldual
algorithm to quantify the uncertainty associated with particular 3D
structures appearing in the maximum a posteriori (MAP) estimate of the
HS cube. We investigate the interest of the proposed method for wideband
radio-interferometric (RI) imaging that consists in inferring the wideband
sky image from incomplete and noisy Fourier measurements. We showcase
the performance of our approach on realistic simulations.
Bayesian uncertainty quantification (UQ) for hyperspectral (HS) inverse
imaging problems. The proposed approach generalizes our recent work
for single-channel UQ [1]. Similarly, the Bayesian hypothesis test is
formulated as a convex minimization problem and solved using a primaldual
algorithm to quantify the uncertainty associated with particular 3D
structures appearing in the maximum a posteriori (MAP) estimate of the
HS cube. We investigate the interest of the proposed method for wideband
radio-interferometric (RI) imaging that consists in inferring the wideband
sky image from incomplete and noisy Fourier measurements. We showcase
the performance of our approach on realistic simulations.
| Original language | English |
|---|---|
| Title of host publication | SPARS 2019 |
| Number of pages | 2 |
| Publication status | Published - 1 Jul 2019 |
| Event | Signal Processing with Adaptive Sparse Structured Representations (SPARS) workshop - Toulouse, France Duration: 1 Jul 2019 → 4 Jul 2019 |
Workshop
| Workshop | Signal Processing with Adaptive Sparse Structured Representations (SPARS) workshop |
|---|---|
| Abbreviated title | SPARS 2019 |
| Country/Territory | France |
| City | Toulouse |
| Period | 1/07/19 → 4/07/19 |