create 3D binary image

Question

I have a 2D array, a, comprising a set of 100 x,y,z coordinates:

[[ 0.81  0.23  0.52]
 [ 0.63  0.45  0.13]
 ...
 [ 0.51  0.41  0.65]]

I would like to create a 3D binary image, b, with 101 pixels in each of the x,y,z dimensions, of coordinates ranging between 0.00 and 1.00. Pixels at locations defined by a should take on a value of 1, all other pixels should have a value of 0.

I can create an array of zeros of the right shape with b = np.zeros((101,101,101)), but how do I assign coordinate and slice into it to create the ones using a?

Answer 1

First, start off by safely rounding your floats to ints. In context, see this question.

a_indices = np.rint(a * 100).astype(int)

Next, assign those indices in b to 1. But be careful to use an ordinary list instead of the array, or else you'll trigger the usage of index arrays. It seems as though performance of this method is comparable to that of alternatives (Thanks @Divakar! :-)

b[list(a_indices.T)] = 1

I created a small example with size 10 instead of 100, and 2 dimensions instead of 3, to illustrate:

>>> a = np.array([[0.8, 0.2], [0.6, 0.4], [0.5, 0.6]])
>>> a_indices = np.rint(a * 10).astype(int)
>>> b = np.zeros((10, 10))
>>> b[list(a_indices.T)] = 1
>>> print(b) 
[[ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  1.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]]