Numpy copy non-zero elements from one array to another (3D array)

Question

my question is if there is an easy way to copy non zero values from one numpy 3d array to another. I wouldn't like to create 3 for loops for that...

Let's say I have an array a:

a = np.array([ [ [1,2,3], [4,5,6]],[[7,8,9], [10,11,12] ] ])
# to visualize it better:
# a = np.array([
#   [  
#     [1,2,3], 
#     [4,5,6]
#   ], 
#   [ 
#     [7,8,9], 
#     [10,11,12] 
#   ] 
# ])
#

then there is an array b:

b = np.array([ [[3,0,9], [0,0,0]], [[0,0,0], [45,46,47]] ])
# to visualize it better:
# b = np.array([
#   [  
#     [3,0,9], 
#     [0,0,0]
#   ], 
#   [ 
#     [0,0,0], 
#     [45,46,47] 
#   ] 
# ])
#

And I would like to merge those arrays to receive non-zero elements from b and other elements from a (these elements that are 0s in b) SO the output would look like:

# 
# np.array([
#   [  
#     [3,2,9], 
#     [4,5,6]
#   ], 
#   [ 
#     [7,8,9], 
#     [45,46,47] 
#   ] 
# ])
#

It doesn't have to be numpy, it can be openCV, but still I would like to know how to achieve this.

Answer 1

You can try using np.where to select from b with the condition b!=0 , or else to select from a :

combined_array = np.where(b!=0, b, a)

>>> combined_array
array([[[ 3,  2,  9],
        [ 4,  5,  6]],

       [[ 7,  8,  9],
        [45, 46, 47]]])

Answer 2

This should do it:

import numpy as np

a = np.array([ [ [1,2,3], [4,5,6]],[[7,8,9], [10,11,12] ] ])
b = np.array([ [[3,0,9], [0,0,0]], [[0,0,0], [45,46,47]] ])

c = b.copy()
c[b==0] = a[b==0]
print(c)

#[[[ 3  2  9]
#  [ 4  5  6]]
#
# [[ 7  8  9]
#  [45 46 47]]]

Where b==0 is an array with the same shape as b where the elements are True if the corresponding element in b equals 0 and False otherwise. You can then use that to select the zero elements of b and replace them with the values at those indices in a .

Edit: The other answer with np.where is nicer.