aug_unsharp_mask — Unsharp spectral masking¶
Group: Augmentation · C ABI: n4m_augmentation_unsharp_mask_*
Description¶
Random unsharp spectral mask.
Parameters¶
Name |
Type |
Default |
|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
API and bindings¶
C ABI (ABI 2): n4m_augmentation_unsharp_mask_apply · n4m_augmentation_unsharp_mask_create · n4m_augmentation_unsharp_mask_destroy. Use the linked public header for the exact signature, configuration, and result handles.
Python (verified public re-export):
from n4m.augmentation.spectral import UnsharpMask
Source signature: UnsharpMask(amount_lo: float = 0.1, amount_hi: float = 0.5, sigma: float = 1.0, kernel_width: int = 11, rng: Optional[PCG64] = None, seed: int = 0).
R: no current source-verified entry point was found for this catalog method.
MATLAB / Octave: no current source-verified entry point was found for this catalog method.
Explanations¶
Bibliographic source¶
Unsharp masking is a classical signal/image sharpening construction, but no single NIR paper defines this stochastic parameterization; source code is normative.
Mathematical principle¶
Build a Gaussian-smoothed row \(S_i=G_{\sigma,w}*X_i\), draw \(a_i\sim U(\text{amount_lo},\text{amount_hi})\), and compute \(X'_i=X_i+a_i(X_i-S_i)\). sigma and odd kernel_width set the low-pass scale; amount_* controls sharpening.
Appropriate uses¶
Varying apparent spectral resolution and sensitivity to narrow bands during training.
Limits and validation¶
Sharpening amplifies high-frequency noise and can create overshoot. Reflect padding changes behavior near the first and last wavelengths.
Implementation¶
Python role API n4m.augmentation.spectral.UnsharpMask; ABI 2 family n4m_augmentation_unsharp_mask_{create,apply,destroy}.
The ABI-2 implementation is the n4m_augmentation_unsharp_mask_* lifecycle in libn4m.
Sources and provenance¶
See also: methods index · interactive dashboard