How do I determine how many windows with overlap fit into an array?

Asked Feb 05 '20 at 15:50

Active Feb 05 '20 at 15:56

Viewed 252 times

The view_as_image function of skimage returns windows of numpy arrays of predetermined size. For example, view_as_windows(array, (20, 96), 5) will return arrays windows of (20, 96) of the original array, sliding 5 rows at a time (if the width of the original array is also 96). I need to determine in advance how many windows will be output by skimage.

Say I have this array:

import numpy as np
from skimage.util.shape import view_as_windows

array = np.random.randint(0, 255, (128, 96))

And I get windows of (20, 96), with a stride of 5.

view_as_windows(array, (20, 96), 5)

It will yield 22 windows. Like this:

first: (0:20, :)
second: (5:25, :)
third: (10:30, :)

How can I do this, i.e. get the number of windows it will create?

edited Feb 05 '20 at 15:56

asked Feb 05 '20 at 15:50

Nicolas Gervais

33,817
13
115
143

You just want the count of windows, i.e. 22? – Divakar Feb 05 '20 at 15:52
yes just that. edited it for clarity – Nicolas Gervais Feb 05 '20 at 15:55
If you just need the count, you can simply do `len(view_as_windows(array, (20, 96), 5))`. `view_as_winodws` would simply be a view, hence no memory overhead, if that's bothering you. – Divakar Feb 05 '20 at 15:56

How do I determine how many windows with overlap fit into an array?

0 Answers0