wassname/scikit-image
1
0
Fork 0
mirror of https://github.com/wassname/scikit-image.git synced 2026-07-05 17:55:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

Replace frame with plane

Browse Source
This commit is contained in:
Juan Nunez-Iglesias
2015-02-03 14:08:24 +11:00
parent f1b309ecfb
commit fd6c7a26d8
1 changed files with 13 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
doc/source/user_guide/numpy_images.txt
+13 -13
View File
@@ -147,7 +147,7 @@ contains the color information and is denoted ``channel`` or ``ch``.
Finally, for 3D images, such as videos, magnetic resonance imaging
(MRI) scans, or confocal microscopy, we refer to the leading dimension
as ``frame``, abbreviated as ``frm`` or ``f``.
as ``plane``, abbreviated as ``pln`` or ``p``.
These conventions are summarized below:
@@ -158,8 +158,8 @@ Image type coordinates
======================== ========================================
2D grayscale (row, col)
2D multichannel (eg. RGB) (row, col, ch)
3D grayscale (frm, row, col)
3D multichannel (frm, row, col, ch)
3D grayscale (pln, row, col)
3D multichannel (pln, row, col, ch)
======================== ========================================
@@ -180,14 +180,14 @@ to process these images:
>>> slics = segmentation.slic(im3d, spacing=[5, 1, 1], multichannel=False)
Other times, processing must be done frame-wise. When frames are the
Other times, processing must be done plane-wise. When planes are the
leading dimension, we can use the following syntax:
>>> from skimage import filters
>>> edges = np.zeros_like(im3d)
>>> for frm, image in enumerate(im3d):
... # iterate over the leading dimension (frames)
... edges[frm] = filters.sobel(image)
>>> for pln, image in enumerate(im3d):
... # iterate over the leading dimension (planes)
... edges[pln] = filters.sobel(image)
Notes on array order
@@ -202,19 +202,19 @@ next to each other in memory is faster than processing them
in a different order, even if the number of operations is the same:
>>> def in_order_multiply(arr, scalar):
... for frame in list(range(arr.shape[0])):
... arr[frame, :, :] *= scalar
... for plane in list(range(arr.shape[0])):
... arr[plane, :, :] *= scalar
...
>>> def out_of_order_multiply(arr, scalar):
... for frame in list(range(arr.shape[2])):
... arr[:, :, frame] *= scalar
... for plane in list(range(arr.shape[2])):
... arr[:, :, plane] *= scalar
...
>>> import time
>>> im3d = np.random.rand(100, 1024, 1024)
>>> t0 = time.time(); x = in_order_multiply(im3d, 5); t1 = time.time()
>>> print("%.2f seconds" % (t1 - t0)) # doctest: +SKIP
0.14 seconds
>>> im3d_t = np.transpose(im3d).copy() # place "frames" dimension at end
>>> im3d_t = np.transpose(im3d).copy() # place "planes" dimension at end
>>> im3d_t.shape
(1024, 1024, 100)
>>> s0 = time.time(); x = out_of_order_multiply(im3d, 5); s1 = time.time()
@@ -237,7 +237,7 @@ A note on time
Although scikit-image does not currently (0.11) provide functions to
work specifically with time-varying 3D data, our compatibility with
numpy arrays allows us to work quite naturally with a 5D array of the
shape (t, frm, row, col, ch):
shape (t, pln, row, col, ch):
>>> for timepoint in image5d: # doctest: +SKIP
... # each timepoint is a 3D multichannel image