Masking API Reference

Mask creation and structural noise extraction.

Module: masking.kernels

Function: create_stage1_mask_kernel

Create single-pixel blind-spot kernel for Stage 1.

Signature:

create_stage1_mask_kernel(center_size=1)

Parameters:

• center_size (int): Size of center True region (≥1)

Returns:

• np.ndarray: Boolean mask kernel

Example:

from autoStructN2V.masking import create_stage1_mask_kernel

mask = create_stage1_mask_kernel(center_size=1)
# Returns 3x3 with False center

---

Function: create_blind_spot_kernel

Create blind-spot kernel (True everywhere except center).

Signature:

create_blind_spot_kernel(kernel_size)

---

Module: masking.structure

Class: StructuralNoiseExtractor

Extract structural binary masks from noisy images using autocorrelation.

Constructor:

StructuralNoiseExtractor(norm_autocorr=True, log_autocorr=True, 
                         crop_autocorr=True, adapt_autocorr=True, 
                         adapt_CB=50.0, adapt_DF=0.95, center_size=15,
                         base_percentile=50, percentile_decay=1.15, 
                         center_ratio_threshold=0.3, use_center_proximity=False,
                         center_proximity_threshold=0.8, keep_center_component_only=True,
                         max_true_pixels=None)

Parameters: See Configuration Reference for full details.

Methods:

extract_mask(noise_patterns, verbose=False)

Extract binary mask from noisy patterns.

Parameters:

• noise_patterns (list/np.ndarray): Noisy images or patches
• verbose (bool): Print extraction details

Returns:

• tuple: (binary_mask, center_square)
  • binary_mask: Boolean mask (e.g., 11x11)
  • center_square: Processed autocorrelation center

Example:

from autoStructN2V.masking import StructuralNoiseExtractor

extractor = StructuralNoiseExtractor(
    center_size=15,
    base_percentile=50,
    max_true_pixels=25,
)

mask, autocorr = extractor.extract_mask(patches, verbose=True)

clear_cache()

Clear cached computation results.

---

Module: masking.utilities

Function: create_full_mask

Create full mask with random placements of single kernel.

Signature:

create_full_mask(single_masking_kernel, patch_size, mask_percentage, verbose=False)

Parameters:

• single_masking_kernel (np.ndarray): Single mask pattern
• patch_size (int): Full mask size
• mask_percentage (float): Target percentage of True pixels (0-100)
• verbose (bool): Print details

Returns:

• tuple: (full_masking_kernel, prediction_kernel)
  • full_masking_kernel: Full mask with pattern placements
  • prediction_kernel: Only center points marked

Example:

from autoStructN2V.masking import create_full_mask

full_mask, pred_kernel = create_full_mask(
    single_masking_kernel=struct_mask,
    patch_size=64,
    mask_percentage=10.0,
    verbose=True
)

---

Function: create_mask_for_training

Convenience function for creating stage-appropriate masks.

Signature:

create_mask_for_training(stage, kernel=None, patch_size=64, 
                         mask_percentage=15.0, **kwargs)

Parameters:

• stage (str): 'stage1' or 'stage2'
• kernel (np.ndarray, optional): Custom kernel
• patch_size (int): Patch size
• mask_percentage (float): Mask percentage
• **kwargs: Additional parameters

Returns:

• tuple: (full_mask, prediction_kernel)

---

See Also:

• Structural Mask Extraction
• Custom Masks Tutorial