Apply thresholding logic to one column of NumPy array

Question

I have a 3D NumPy array which is essentially image data in RGBA format like so:

[
  [
    [233 228 230 120]
    [233 228 230 0]
    [232 227 229 212]
    ...
  ]
  ...
]

In the example above the last (4th) column represents alpha channel. As I can't have semi-transparent pixels (value other than 0 or 255), I need to threshold that column. Any value below 255 should become 0.

What I have is the line below.

image[...,3][image[...,3] < 255] = 0

It does work but I wanted to ask if there is something shorter and more elegant that would do the job. Thanks.

can you post your image shape?

sai
– sai

2020-12-15 21:40:48 +00:00
Commented Dec 15, 2020 at 21:40 — sai
– sai, Commented Dec 15, 2020 at 21:40

sai · Accepted Answer · 2020-12-15 22:49:50Z

1

Edit: @Lior Cohen was indeed right! Apologies.

# if your channels are in the last axis, this should work
image[:,:,3][image[:,:,3] < 255] = 0

# another alternative, but not as elegant as what @Lior proposed
image[:,:,3] = np.where(image[:,:,3] < 255, 0, 255)

edited Dec 15, 2020 at 22:49

answered Dec 15, 2020 at 21:42

sai

1,8041 gold badge9 silver badges13 bronze badges

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

2 Comments

Lior Cohen Over a year ago

I think this will truncate the RGB to 0 and not only the alpha

sai Over a year ago

@LiorCohen you were right! I overlooked it. Your answer should be accepted.

Lior Cohen · Accepted Answer · 2020-12-15 22:29:33Z

1

Two More options that are based on the /255 operator which will cut integer lower than 255 to 0.

image[...,3] = image[...,3] // 255 * 255

image[...,3] //= 255
image[...,3] *= 255

answered Dec 15, 2020 at 22:29

Lior Cohen

5,7202 gold badges18 silver badges33 bronze badges

Collectives™ on Stack Overflow

Apply thresholding logic to one column of NumPy array

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related