plenoptic.process.expand

plenoptic.process.expand(x, factor)

Expand a signal by a factor.

We do this in the frequency domain: pasting the Fourier contents of x in the center of a larger empty tensor, and then taking the inverse FFT.

Parameters:

• x (Tensor) – The signal for expansion.

• factor (float) – Factor by which to resize image. Must be larger than 1 and factor * x.shape[-2:] must give integer values.

Return type:

Tensor

Returns:

expanded – The expanded signal.

Raises:

• ValueError – If factor is less than or equal to 1.

• ValueError – If factor times the height or width of x is not an integer.

See also

shrink

The inverse operation.

upsample_blur

An alternative upsampling operation.

Examples

>>> import plenoptic as po
>>> import torch
>>> img = po.data.einstein()
>>> img.shape
torch.Size([1, 1, 256, 256])
>>> expanded = po.process.expand(img, factor=2)
>>> expanded.shape
torch.Size([1, 1, 512, 512])
>>> po.plot.imshow(
...     [img, expanded], title=["original", "expanded"], zoom=0.5, vrange=(0, 1)
... )
<PyrFigure...>

(png, hires.png, pdf)

Note that the range has changed:

>>> img.min(), img.max()
(tensor(0.0039), tensor(1.))
>>> expanded.min(), expanded.max()
(tensor(-0.1648), tensor(1.0239))

An alternative method for upsampling images is to use upsample_blur:

>>> po.plot.imshow(
...     [img, expanded, po.process.upsample_blur(img, (0, 0))],
...     title=["original", "expanded", "blurred"],
...     vrange=(0, 1),
... )
<PyrFigure...>

(png, hires.png, pdf)