In [1]:

    

import numpy as np


    

In [2]:

    

import skimage.util


    

In [3]:

    

a = np.arange(24).reshape(4, 6)
print(a)


    

    




[[ 0  1  2  3  4  5]
 [ 6  7  8  9 10 11]
 [12 13 14 15 16 17]
 [18 19 20 21 22 23]]

In [4]:

    

blocks = skimage.util.view_as_blocks(a, (2, 3))
print(blocks)


    

    




[[[[ 0  1  2]
   [ 6  7  8]]

  [[ 3  4  5]
   [ 9 10 11]]]


 [[[12 13 14]
   [18 19 20]]

  [[15 16 17]
   [21 22 23]]]]

In [5]:

    

print(type(blocks))


    

    




<class 'numpy.ndarray'>

In [6]:

    

print(blocks.shape)


    

    




(2, 2, 2, 3)

In [7]:

    

print(blocks[0, 0])


    

    




[[0 1 2]
 [6 7 8]]

In [8]:

    

print(blocks[0, 1])


    

    




[[ 3  4  5]
 [ 9 10 11]]

In [9]:

    

print(blocks[1, 0])


    

    




[[12 13 14]
 [18 19 20]]

In [10]:

    

print(blocks[1, 1])


    

    




[[15 16 17]
 [21 22 23]]

In [11]:

    

# blocks = skimage.util.view_as_blocks(a, (2, 4))
# ValueError: 'block_shape' is not compatible with 'arr_in'


    

In [12]:

    

print(np.shares_memory(a, blocks))


    

    




True

In [13]:

    

a[0, 0] = 100


    

In [14]:

    

print(blocks[0, 0])


    

    




[[100   1   2]
 [  6   7   8]]

In [15]:

    

a = np.arange(24).reshape(4, 6)
blocks_copy = skimage.util.view_as_blocks(a, (2, 3)).copy()


    

In [16]:

    

print(np.shares_memory(a, blocks_copy))


    

    




False

In [17]:

    

a[0, 0] = 100


    

In [18]:

    

print(blocks_copy[0, 0])


    

    




[[0 1 2]
 [6 7 8]]

In [19]:

    

blocks_copy2 = skimage.util.view_as_blocks(a.copy(), (2, 3))


    

In [20]:

    

print(np.shares_memory(a, blocks_copy2))


    

    




False