Mask 3d numpy array

Question

Consider the following array as an image with three layers (red, green, blue). For each position in the image i want to test whether there is a specific value in one of the layers (e.g. 99). If that is true the values at the same position in the other layers shall be replaced by this specific value.

img = np.array([[[0, 99],
                 [2, 3]],
                     
                 [[4, 5],
                  [6, 7]],
                     
                 [[8, 9],
                  [10, 11]]])

The desired result would be:

img = [[[0, 99],
        [2, 3]],

       [[4, 99],
        [6, 7]],

       [[8, 99],
        [10, 11]]]

I am currently using the following code to fulfill the task.

c= []
for i in range(img.shape[1]):
    for j in range(img.shape[2]):
        if np.any(img[:, i, j]==99):
            c.append([99 for x in img[:, i, j]])
        else: c.append(img[:, i, j])

new_img = [np.array([row[i] for row in c]).reshape((2,2)).tolist() for i in range(len(c))]

However, this seems somewhat complicated. Can you recommend a better approach? Thank you for your help.

@MichaelSzczesny. Sorry, didn't see your comment when I was posting. — Mad Physicist
– Mad Physicist, Commented Jun 9, 2022 at 21:43
@MichaelSzczesny. Fair enough. I bothered answering because it's the first one I recall seeing where the masked axes are not the leading axes. — Mad Physicist
– Mad Physicist, Commented Jun 9, 2022 at 21:46

Mad Physicist · Accepted Answer · 2022-06-09 21:42:48Z

2

The mask of pixels that are 99 in any channel is

mask = (img == 99).any(0)

You can apply this mask directly:

img[:, mask] = 99

Or as a one-liner:

img[:, (img == 99).any(0)] = 99

answered Jun 9, 2022 at 21:42

Mad Physicist

116k29 gold badges202 silver badges292 bronze badges

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

Comments

Ali_Sh · Accepted Answer · 2022-06-09 21:45:47Z

0

You can do this by:

ind = np.argwhere(img == 99)[0, 1:]   # [0, 1]
img[:, ind[0], ind[1]] = 99
# [[[ 0 99]
#   [ 2  3]]
#  [[ 4 99]
#   [ 6  7]]
#  [[ 8 99]
#   [10 11]]]

which will work just on the first occurrence of 99. So, if you have more than one layer contain 99 or more than one count of 99 in each layer, do as @Mad solution.

edited Jun 9, 2022 at 21:45

answered Jun 9, 2022 at 21:37

Ali_Sh

2,8364 gold badges45 silver badges70 bronze badges

Collectives™ on Stack Overflow

Mask 3d numpy array

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related