"""Module for calculating the multiscale structural similarity index (MS-SSIM) between
two images.
Examples
--------
.. doctest-skip::
>>> import numpy as np
>>> from viqa import MSSSIM
>>> img_r = np.random.rand(256, 256)
>>> img_m = np.random.rand(256, 256)
>>> msssim = MSSSIM()
>>> msssim.score(img_r, img_m, data_range=1)
"""
# Authors
# -------
# Author: Lukas Behammer
# Research Center Wels
# University of Applied Sciences Upper Austria, 2023
# CT Research Group
#
# Modifications
# -------------
# Original code, 2024, Lukas Behammer
#
# License
# -------
# BSD-3-Clause License
from warnings import warn
import numpy as np
from piq import multi_scale_ssim
from torch import tensor
from viqa._metrics import FullReferenceMetricsInterface
from viqa.utils import _check_chromatic
[docs]
class MSSSIM(FullReferenceMetricsInterface):
"""Calculate the multiscale structural similarity index (MS-SSIM) between two
images.
Attributes
----------
score_val : float
MS-SSIM score value of the last calculation.
parameters : dict
Dictionary containing the parameters for MS-SSIM calculation.
Parameters
----------
data_range : {1, 255, 65535}, default=255
Data range of the returned data in data loading. Is used for image loading when
``normalize`` is True and for the MS-SSIM calculation. Passed to
:py:func:`viqa.utils.load_data` and :py:meth:`score`.
normalize : bool, default=False
If True, the input images are normalized to the ``data_range`` argument.
**kwargs : optional
Additional parameters for data loading. The keyword arguments are passed to
:py:func:`.viqa.utils.load_data`.
Other Parameters
----------------
chromatic : bool, default False
If True, the input images are expected to be RGB images.
.. note::
Color images can be used, but it is unclear how the called implementation
:py:func:`piq.multi_scale_ssim` handles the color channels.
Raises
------
ValueError
If ``data_range`` is not set.
Warnings
--------
This metric is not yet tested. The metric should be only used for experimental
purposes.
.. todo:: test
Notes
-----
For more information on the MS-SSIM metric, see [1]_.
See Also
--------
viqa.fr_metrics.uqi.UQI : Universal quality index (UQI) between two images.
viqa.fr_metrics.ssim.SSIM : Structural similarity index (SSIM) between two images.
References
----------
.. [1] Wang, Z., Simoncelli, E. P., & Bovik, A. C. (2003). Multi-scale structural
similarity for image quality assessment. The Thirty-Seventh Asilomar Conference
on Signals, Systems & Computers, 1298–1402.
https://doi.org/10.1109/ACSSC.2003.1292216
"""
def __init__(self, data_range=255, normalize=False, **kwargs):
"""Construct method."""
if data_range is None:
raise ValueError("Parameter data_range must be set.")
super().__init__(data_range=data_range, normalize=normalize, **kwargs)
self._name = "MS-SSIM"
[docs]
def score(self, img_r, img_m, dim=None, im_slice=None, **kwargs):
"""
Calculate the multiscale structural similarity index (MS-SSIM) between two
images.
Parameters
----------
img_r : np.ndarray or Tensor or str or os.PathLike
Reference image to calculate score against.
img_m : np.ndarray or Tensor or str or os.PathLike
Distorted image to calculate score of.
dim : {0, 1, 2}, optional
MS-SSIM for 3D images is calculated as mean over all slices of the given
dimension.
im_slice : int, optional
If given, MS-SSIM is calculated only for the given slice of the 3D image.
**kwargs : optional
Additional parameters for MS-SSIM calculation. The keyword arguments are
passed to :py:func:`piq.multi_scale_ssim`. See the documentation under [2]_.
Other Parameters
----------------
kernel_size : int, default=11
The side-length of the sliding window used in comparison. Must be an odd
value.
kernel_sigma : float, default=1.5
Sigma of normal distribution.
reduction : str, default='mean'
Specifies the reduction type: 'none', 'mean' or 'sum'.
scale_weights : list, default=[0.0448, 0.2856, 0.3001, 0.2363, 0.1333]
Weights for different scales.
k1 : float, default=0.01
Algorithm parameter, K1 (small constant, see [3]_).
k2 : float, default=0.03
Algorithm parameter, K2 (small constant, see [3]_).
Try a larger K2 constant (e.g. 0.4) if you get a negative or NaN
results.
.. seealso::
See :py:func:`.viqa.fr_metrics.ssim.structural_similarity` for more
information
Returns
-------
score_val : float
MS-SSIM score value.
Raises
------
ValueError
If invalid dimension given in ``dim``. \n
If images are neither 2D nor 3D. \n
If images are 3D, but dim is not given. \n
If ``im_slice`` is given, but not an integer.
Warns
-----
RuntimeWarning
If ``dim`` or ``im_slice`` is given for 2D images. \n
If ``im_slice`` is not given, but ``dim`` is given for 3D images, MS-SSIM is
calculated for the full volume.
Notes
-----
For 3D images if ``dim`` is given, but ``im_slice`` is not, the MS-SSIM is
calculated for the full volume of the 3D image. This is implemented as `mean` of
the MS-SSIM values of all slices of the given dimension. If ``dim`` is given and
``im_slice`` is given, the MS-SSIM is calculated for the given slice of the
given dimension (represents a 2D metric of the given slice).
References
----------
.. [2] https://piq.readthedocs.io/en/latest/functions.html#piq.multi_scale_ssim
.. [3] Wang, Z., Simoncelli, E. P., & Bovik, A. C. (2003). Multi-scale
structural similarity for image quality assessment. The Thirty-Seventh
Asilomar Conference on Signals, Systems & Computers, 1298–1402.
https://doi.org/10.1109/ACSSC.2003.1292216
"""
img_r, img_m = self.load_images(img_r, img_m)
if "scale_weights" in kwargs and type(kwargs["scale_weights"]) is list:
kwargs["scale_weights"] = tensor(kwargs["scale_weights"])
if img_r.ndim == 3 and img_r.shape[-1] != 3:
if (
dim is not None and type(im_slice) is int
): # if dim and im_slice are given
# Calculate MS-SSIM for given slice of given dimension
match dim:
case 0:
img_r_tensor, img_m_tensor = _check_chromatic(
img_r[im_slice, :, :],
img_m[im_slice, :, :],
self.parameters["chromatic"],
)
score_val = multi_scale_ssim(
img_r_tensor,
img_m_tensor,
data_range=self.parameters["data_range"],
**kwargs,
)
case 1:
img_r_tensor, img_m_tensor = _check_chromatic(
img_r[:, im_slice, :],
img_m[:, im_slice, :],
self.parameters["chromatic"],
)
score_val = multi_scale_ssim(
img_r_tensor,
img_m_tensor,
data_range=self.parameters["data_range"],
**kwargs,
)
case 2:
img_r_tensor, img_m_tensor = _check_chromatic(
img_r[:, :, im_slice],
img_m[:, :, im_slice],
self.parameters["chromatic"],
)
score_val = multi_scale_ssim(
img_r_tensor,
img_m_tensor,
data_range=self.parameters["data_range"],
**kwargs,
)
case _:
raise ValueError(
"Invalid dim value. Must be integer of 0, 1 or 2."
)
elif (
dim is not None and im_slice is None
): # if dim is given, but im_slice is not, calculate MS-SSIM for full
# volume
warn(
"im_slice is not given. Calculating MS-SSIM for full volume.",
RuntimeWarning,
)
img_r_tensor, img_m_tensor = _check_chromatic(
img_r,
img_m,
self.parameters["chromatic"],
)
score_val = multi_scale_ssim(
img_r_tensor,
img_m_tensor,
data_range=self.parameters["data_range"],
**kwargs,
)
else:
if type(im_slice) is not int or None:
raise ValueError("im_slice must be an integer.")
raise ValueError(
"If images are 3D, dim and im_slice (optional) must be given."
)
elif img_r.ndim == 2 or (img_r.ndim == 3 and img_r.shape[-1] == 3):
if dim or im_slice:
warn("dim and im_slice are ignored for 2D images.", RuntimeWarning)
# Calculate MS-SSIM for 2D images
img_r_tensor, img_m_tensor = _check_chromatic(
img_r,
img_m,
self.parameters["chromatic"],
)
score_val = multi_scale_ssim(
img_r_tensor,
img_m_tensor,
data_range=self.parameters["data_range"],
**kwargs,
)
else:
raise ValueError("Images must be 2D or 3D.")
self.score_val = float(score_val)
return score_val
[docs]
def print_score(self, decimals=2):
"""Print the MSSSIM score value of the last calculation.
Parameters
----------
decimals : int, default=2
Number of decimal places to print the score value.
Warns
-----
RuntimeWarning
If :py:attr:`score_val` is not available.
"""
if self.score_val is not None:
print("MS-SSIM: {}".format(np.round(self.score_val, decimals)))
else:
print("No score value for MS-SSIM. Run score() first.")