Table 1
Comparison of assumptions made in the PCA, I-PCA, REXPACO, MAYO, and mustard algorithms with respect to reality.
| PCA | I-PCA | REXPACO | MAYO | mustard | Reality | |
|---|---|---|---|---|---|---|
| Speckle field | PCs(a) | Positive, PCs | Positive, static | Positive, PCs from I-PCA | Positive, static | Positive, quasi-static |
| Disk signal | Rotating | Positive, rotating | Positive, rotating | Positive, rotating, shearlets | Positive, rotating | Positive, rotating |
| Noise | N/A | N/A | Learned statistics, by patches, from background | Constrained (b), white noise | White noise (implicit)(c) | White noise, shot noise, Rician speckle noise, + others (d) |
| Flux invariant to rotation | All belong to speckle field | All belong to speckle field | All belong to speckle field | All belong to speckle field | Sorted according to Gaussian mask(e) | A fraction belongs to the disk |
Notes. (a)Principal component(s). Using one PC means that the estimated speckle field can vary in amplitude, while more components will capture variation in its morphology. (b)Noise statistics learned from I-PCA results and enforced within Huber loss. (c)Scope of validity of MSE estimator. (d)Other nuisance phenomena, which create temporally varying extended structures, such as wind-driven halos (Cantalloube et al. 2019), can be accounted for noise.(e)Mreg drives the rotation-invariant flux assignment (see Sect. 3.1).
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.