spyrit.core.prep.SplitPoisson.sigma_expe

SplitPoisson.sigma_expe(x: tensor) → tensor

Estimates the variance of the measurements that are compensated for splitting but NOT for image normalization

Args:

x: Batch of images in the Hadamard domain.

Shape:

Input: \((B,2*M)\) where \(B\) is the batch dimension

Output: \((B, M)\)

Example:

>>> x = torch.rand([10,2*32*32], dtype=torch.float)
>>> split_op.set_expe(gain=1.6)
>>> v = split_op.sigma_expe(x)
>>> print(v.shape)
torch.Size([10, 400])