mrpro.utils.filters.gaussian_filter

mrpro.utils.filters.gaussian_filter(x: Tensor, sigmas: float | Sequence[float] | Tensor, dim: int | Sequence[int] | None = None, truncate: int = 3, pad_mode: Literal['constant', 'reflect', 'replicate', 'circular'] = 'constant', pad_value: float = 0.0) → Tensor[source]

Apply a and-Gaussian filter.

Parameters:

  • x – Tensor to filter

  • sigmas – Standard deviation for Gaussian kernel. If iterable, must have length equal to the number of axes.

  • dim – Axis or axes to filter over. If None, filters over all axes.

  • truncate – Truncate the filter at this many standard deviations.

  • pad_mode – Padding mode

  • pad_value – Padding value for pad_mode = constant