pado.math.gaussian_window

gaussian_window(size, sigma, *, device=None, dtype=torch.float32)[source]

Create normalized 2D Gaussian window.

Parameters:
  • size (int) – Width and height of square window

  • sigma (float) – Standard deviation of Gaussian

  • device (torch.device, optional) – Device for direct kernel construction

  • dtype (torch.dtype) – Real floating-point kernel dtype (default float32)

Returns:

Normalized 2D Gaussian window [size, size]

Return type:

torch.Tensor

Examples

>>> window = gaussian_window(11, 1.5)