neu_vol.backends.view

Read-only view backends that wrap another backend.

CropBackend is a shifted/resized window over a source, padding out-of-bounds regions with a fill value; MaskBackend blanks boxes to a fill value; RemapBackend replaces values. Each opens from a spec, so the engine can crop, pad, mask and pyramid a region as it would any other source.

Masking on the read side holds at every scale: each pyramid level is derived from the level below it in the output, so a hole in level 0 propagates upward on its own.

Attributes

MASK_TAG
REMAP_TAG
TAG

Classes

CropBackend	A window [origin, origin+shape) over a source backend (canonical order).
MaskBackend	A source with every box in boxes replaced by value (canonical order).
RemapBackend	A source with values replaced by to (default 0) as it is read.

Module Contents

class neu_vol.backends.view.CropBackend(spec: Mapping[str, Any])

A window [origin, origin+shape) over a source backend (canonical order).

origin may be negative and shape may extend past the source; regions outside the source are filled with pad_value.

classmethod open(spec: Mapping[str, Any]) → CropBackend

read_region(region: neu_vol.backends.base.Region) → numpy.ndarray

to_spec() → dict[str, Any]

write_region(region: neu_vol.backends.base.Region, data: numpy.ndarray) → None

property chunks: tuple[int, ...]

property dtype: numpy.dtype

property shape: tuple[int, ...]

class neu_vol.backends.view.MaskBackend(spec: Mapping[str, Any])

A source with every box in boxes replaced by value (canonical order).

Boxes are full-ndim half-open [lo, hi) in the source’s own index space, so a mask composes with a crop laid over it without either knowing about the other: the crop shifts coordinates, and this sits underneath.

classmethod open(spec: Mapping[str, Any]) → MaskBackend

read_region(region: neu_vol.backends.base.Region) → numpy.ndarray

to_spec() → dict[str, Any]

write_region(region: neu_vol.backends.base.Region, data: numpy.ndarray) → None

property chunks: tuple[int, ...]

property dtype: numpy.dtype

property shape: tuple[int, ...]

class neu_vol.backends.view.RemapBackend(spec: Mapping[str, Any])

A source with values replaced by to (default 0) as it is read.

For a segmentation whose background is not 0: a block of background 1s is not all-fill, so it gets stored, and “which chunks exist” stops answering “where is the data”, which occupancy, relabel and --sparse rely on. Remapping on the read side corrects it before the storage decision.

classmethod open(spec: Mapping[str, Any]) → RemapBackend

read_region(region: neu_vol.backends.base.Region) → numpy.ndarray

stored_axes(name: str = 'axes')

stored_offset(name: str = 'voxel_offset')

stored_voxel_size(name: str = 'voxel_size')

to_spec() → dict[str, Any]

write_region(region: neu_vol.backends.base.Region, data: numpy.ndarray) → None

property chunks: tuple[int, ...]

property dtype: numpy.dtype

property shape: tuple[int, ...]

neu_vol.backends.view.MASK_TAG = 'mask'

neu_vol.backends.view.REMAP_TAG = 'remap'

neu_vol.backends.view.TAG = 'crop'