Importing image from array and axis reorder

matt.mccormick · August 10, 2018, 2:00pm

ITK indices, by convention, are [i,j,k] while NumPy indices are [k,j,i] by convention.

That is

itk_image.GetPixel([i,j,k])

is the same as

numpy_array_as_image[k,j,i]

With ITK 4.13.1, the keepAxes keyword argument to GetArrayFromImage can be used to reorder the array; in ITK 5, this will be keep_axes as we become more PEP8 compliant / Pythonic with the Python API. As noted in the documentation, the numpy operator here for GetImageFromArray to flip the image is numpy.reshape:

github.com

InsightSoftwareConsortium/ITK/blob/efb7fcb84f8afa835bafce729a9faf4b4c4e2242/Modules/Bridge/NumPy/wrapping/PyBuffer.i.in#L8-L19


def GetArrayViewFromImage(image, keepAxes=False, updateLargestPossibleRegion=True):
    """Get a NumPy array view of a ITK Image.


    When *keepAxes* is *False*, the NumPy array will have C-order
    indexing. This is the reverse of how indices are specified in ITK,
    i.e. k,j,i versus i,j,k. However C-order indexing is expected by most
    algorithms in NumPy / SciPy.


    Warning: No copy of the data is performed. Using an array
    view after its source image has been deleted can results in corrupt values
    or a segfault.
    """

github.com

InsightSoftwareConsortium/ITK/blob/efb7fcb84f8afa835bafce729a9faf4b4c4e2242/Modules/Bridge/NumPy/wrapping/PyBuffer.i.in#L123-L140


"""Get an ITK Image of a NumPy array.


This is a deep copy of the NumPy array buffer and is completely safe without potential
side effects.


If isVector is True, then a 3D array will be treated as a 2D vector image,
otherwise it will be treated as a 3D image.


If the array uses Fortran-order indexing, i.e. i,j,k, the Image Size
will have the same dimensions as the array shape. If the array uses
C-order indexing, i.e. k,j,i, the image Size will have the dimensions
reversed from the array shape.


Therefore, since the *numpy.transpose* operator on a 2D array simply
inverts the indexing scheme, the Image representation will be the
same for an array and its transpose. If flipping is desired, see
*numpy.reshape*.
"""

However, it is strongly recommended not to do this, because images will be flipped in matplotlib, etc. and it has memory usage and performance downsides.