normalize_preds#

from pssr.util import normalize_preds
pssr.util.normalize_preds(hr: ndarray, hr_hat: ndarray, pmin: float = 0.1, pmax: float = 99.9)#

Normalizes prediction image intensities to ground truth for fair benchmarking.

Parameters:

  • hr (ndarray) – High-resolution ground truth images as array.

  • hr_hat (ndarray) – High-resolution prediction images as array.

  • pmin (float) – Percentile minimum image intensity. Default is 0.1.

  • pmax (float) – Percentile maximum image intensity. Default is 99.9.

Returns:

Normalized high-resolution ground truth image.

hr_hat_norm (ndarray) : Normalized high-resolution prediction image.

Return type:

hr_norm (ndarray)