Python element-wise condition on numpy ndarray [duplicate]

Question

I have two numpy ndarrays of the same shape.

A = [[12, 25, 6],
    [28, 52, 74]]
B = [[100, 2, 4],
    [2, 12, 14]]

My goal is to replace every element where there the value in B is <= 5 by 0 in A. So my result should be :

# So C[0][0] = 12 because A[0][0] = 12 and B[0][0] >= 5
C = [[12, 0, 0],
    [0, 52, 74]]

Is there an efficient way to do this? For context, this is to try to do some background substraction on images, and replace all background by black color.

np.where(B>5, A, 0)?

not_speshal
– not_speshal

2021-10-19 15:27:15 +00:00
Commented Oct 19, 2021 at 15:27 — not_speshal
– not_speshal, Commented Oct 19, 2021 at 15:27

Ch3steR · Accepted Answer · 2021-10-19 15:28:07Z

3

Here you go:

A = np.array([[12, 25, 6],[28, 52, 74]])
B = np.array([[100, 2, 4],[2, 12, 14]])

A = np.where(B <= 5, 0, A)

Output:

array([[12,  0,  0],
       [ 0, 52, 74]])

edited Oct 19, 2021 at 15:28

Ch3steR

20.8k4 gold badges33 silver badges66 bronze badges

answered Oct 19, 2021 at 15:25

sehan2

1,8551 gold badge14 silver badges27 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Matt Hall Over a year ago

I think this is reasonable if you're creating a new array, but if you're going to clobber A then A[B <= 5] = 0 seems neater. But then I find np.where() a bit unintuitive.

Matt Hall · Accepted Answer · 2021-10-19 15:32:22Z

0

If you want a new array, I would do this:

C = A.copy()
C[B <= 5] = 0

It's a bit faster than np.where() on my machine anyway.

If you don't mind overwriting A, just do A[B <= 5] = 0.

answered Oct 19, 2021 at 15:32

Matt Hall

8,1691 gold badge28 silver badges38 bronze badges

Collectives™ on Stack Overflow

Python element-wise condition on numpy ndarray [duplicate]

2 Answers 2

1 Comment

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Linked

Related