fiery.distmap

Euclidean distance transform in PyTorch.

Functions:

l1_distance_transform

l1_distance_transform(x: Tensor, ndim: int | None = None, vx: float | Sequence[float] = 1) -> Tensor

Compute the L1 distance transform of a binary image

Parameters:

Name	Type	Description	Default
`x`	`(..., *spatial) tensor`	Input tensor. Zeros will stay zero, and the distance will be propagated into nonzero voxels.	required
`ndim`	`int`	Number of spatial dimensions	`x.dim()`
`vx`	`[sequence of] float`	Voxel size	`1`

Returns:

Name	Type	Description
`d`	`(..., *spatial) tensor`	Distance map

References

..[1] "Distance Transforms of Sampled Functions" Pedro F. Felzenszwalb & Daniel P. Huttenlocher Theory of Computing (2012) https://www.theoryofcomputing.org/articles/v008a019/v008a019.pdf

l1_signed_transform

l1_signed_transform(x: Tensor, ndim: int | None = None, vx: float | Sequence[float] = 1) -> Tensor

Compute the signed L1 distance transform of a binary image

Parameters:

Name	Type	Description	Default
`x`	`(..., *spatial) tensor`	Input tensor. A negative distance will propagate into zero voxels and a positive distance will propagate into nonzero voxels.	required
`ndim`	`int`	Number of spatial dimensions	`x.dim()`
`vx`	`[sequence of] float`	Voxel size	`1`

Returns:

Name	Type	Description
`d`	`(..., *spatial) tensor`	Signed distance map

References

..[1] "Distance Transforms of Sampled Functions" Pedro F. Felzenszwalb & Daniel P. Huttenlocher Theory of Computing (2012) https://www.theoryofcomputing.org/articles/v008a019/v008a019.pdf

euclidean_distance_transform

euclidean_distance_transform(x: Tensor, ndim: int | None = None, vx: float | Sequence[float] = 1, squared: bool = False) -> Tensor

Compute the Euclidean distance transform of a binary image

Parameters:

Name	Type	Description	Default
`x`	`(..., *spatial) tensor`	Input tensor. Zeros will stay zero, and the distance will be propagated into nonzero voxels.	required
`ndim`	`int`	Number of spatial dimensions	`x.dim()`
`vx`	`[sequence of] float`	Voxel size	`1`
`squared`	`bool`	Return the squared distance map, skipping the final square root.	`False`

Returns:

Name	Type	Description
`d`	`(..., *spatial) tensor`	Distance map (or squared distance map, if `squared=True`)

References

..[1] "Distance Transforms of Sampled Functions" Pedro F. Felzenszwalb & Daniel P. Huttenlocher Theory of Computing (2012) https://www.theoryofcomputing.org/articles/v008a019/v008a019.pdf

euclidean_signed_transform

euclidean_signed_transform(x: Tensor, ndim: int | None = None, vx: float | Sequence[float] = 1, squared: bool = False) -> Tensor

Compute the signed Euclidean distance transform of a binary image

Parameters:

Name	Type	Description	Default
`x`	`(..., *spatial) tensor`	Input tensor. A negative distance will propagate into zero voxels and a positive distance will propagate into nonzero voxels.	required
`ndim`	`int`	Number of spatial dimensions	`x.dim()`
`vx`	`[sequence of] float`	Voxel size	`1`
`squared`	`bool`	Return the squared distance map, skipping the final square root.	`False`

Returns:

Name	Type	Description
`d`	`(..., *spatial) tensor`	Signed distance map (or squared signed distance map, if `squared=True`)

References

..[1] "Distance Transforms of Sampled Functions" Pedro F. Felzenszwalb & Daniel P. Huttenlocher Theory of Computing (2012) https://www.theoryofcomputing.org/articles/v008a019/v008a019.pdf