nanopyx.methods.channel_registration 

 1from .estimator import ChannelRegistrationEstimator
 2from .corrector import ChannelRegistrationCorrector
 3from ...core.utils.timeit import timeit
 4
 5@timeit
 6def estimate_channel_registration(image_array, ref_channel, max_shift, blocks_per_axis, min_similarity, method="subpixel",
 7                                  save_translation_masks=True, translation_mask_save_path=None, algorithm="field",
 8                                  save_ccms=False, ccms_save_path=False, apply=True):
 9    """
10    Function used to estimate shift between different color channels and align them of an image based on cross correlation.
11    :param image_array:numpy array  with shape (n_channels, y, x); image to be corrected
12    :param ref_channel: int; channel index to be used as reference
13    :param max_shift: int; maximum shift accepted for correction, in pixels.
14    :param blocks_per_axis: int; number of blocks to divide the image in both x and y dimensions
15    :param min_similarity: float; minimum value of similarity to accept a shift as a correction
16    :param method: str; "subpixel" (default) or "max"; subpixel uses a minimizer to find the maximum correlation with
17    subpixel precision, max simply takes the maximum of the cross correlation map
18    :param save_translation_masks: bool, defaults to True; whether to save translation masks as a tif or not
19    :param translation_mask_save_path: str; path where to save translation masks
20    :param save_ccms: bool, defaults to True; whether to save cross correlation matrices as a tif or not
21    :param ccms_save_path: str; path where to save cross correlation matrices
22    :param apply: bool; whether to apply the correction if True or only estimate if False
23    :return: if apply==True, returns corrected image with shape (c, y, x)
24    """
25    estimator = ChannelRegistrationEstimator()
26    aligned_image = estimator.estimate(image_array, ref_channel, max_shift, blocks_per_axis, min_similarity, method=method,
27                                       save_translation_masks=save_translation_masks, translation_mask_save_path=translation_mask_save_path,
28                                       save_ccms=save_ccms, ccms_save_path=ccms_save_path, algorithm=algorithm, apply=apply)
29
30    if aligned_image is not None:
31        return aligned_image
32    else:
33        pass
34
35@timeit
36def apply_channel_registration(image_array, translation_masks=None):
37    """
38    Function used to align different color channels of an image based on cross correlation.
39    :param image_array: numpy array with shape (n_channels, y, x); image to be registered
40    :param translation_masks: numpy array of translation masks
41    :return: returns corrected image with shape (c, y, x)
42    """
43    corrector = ChannelRegistrationCorrector()
44    aligned_image = corrector.align_channels(image_array, translation_masks=translation_masks)
45
46    return aligned_image
def estimate_channel_registration(*args, **kwargs): 
 8    def wrapper(*args, **kwargs):
 9        t = time.time()
10        retval = func(*args, **kwargs)
11        print(f"{func.__name__} took {round(time.time()-t,3)} seconds")
12        return retval

Function used to estimate shift between different color channels and align them of an image based on cross correlation.

Parameters
  • image_array: numpy array with shape (n_channels, y, x); image to be corrected
  • ref_channel: int; channel index to be used as reference
  • max_shift: int; maximum shift accepted for correction, in pixels.
  • blocks_per_axis: int; number of blocks to divide the image in both x and y dimensions
  • min_similarity: float; minimum value of similarity to accept a shift as a correction
  • method: str; "subpixel" (default) or "max"; subpixel uses a minimizer to find the maximum correlation with subpixel precision, max simply takes the maximum of the cross correlation map
  • save_translation_masks: bool, defaults to True; whether to save translation masks as a tif or not
  • translation_mask_save_path: str; path where to save translation masks
  • save_ccms: bool, defaults to True; whether to save cross correlation matrices as a tif or not
  • ccms_save_path: str; path where to save cross correlation matrices
  • apply: bool; whether to apply the correction if True or only estimate if False
Returns

if apply==True, returns corrected image with shape (c, y, x)

def apply_channel_registration(*args, **kwargs): 
 8    def wrapper(*args, **kwargs):
 9        t = time.time()
10        retval = func(*args, **kwargs)
11        print(f"{func.__name__} took {round(time.time()-t,3)} seconds")
12        return retval

Function used to align different color channels of an image based on cross correlation.

Parameters
  • image_array: numpy array with shape (n_channels, y, x); image to be registered
  • translation_masks: numpy array of translation masks
Returns

returns corrected image with shape (c, y, x)