How to programmatically fill in RGB noise in the transparent region of an image using Python?

Question

I need to process a lot of images using Python. All these images have some transparent region (alpha channel) of different sizes.

I need to programmatically fill in RGB noise in the transparent region of those images, but keep the non-transparent region unchanged. This is an example of changing the images.

How to do this programmatically in Python?

Make a new image completely filled with random R,G,B values. Blend the image in question over the top of the random one. — Mark Ransom, Sep 12 '14 at 14:22
slow but simple: `for each pixel: if alpha equals zero: set rgb value to random value and set alpha to 255` — Micka, Sep 12 '14 at 14:25

score 1 · Accepted Answer · answered Sep 12 '14 at 14:22

1

In my opinion you need to:

Create a Mat that contains Gaussian noise (or what kind of noise you need to add in the images).
For each image you copy the noise Mat into another one based on the alpha channel (used as mask)
Add the two images (initial and noise_mask) to he initial image (or the inital_noisy_background)

answered Sep 12 '14 at 14:22

sop

3,445
8
41
84

1

Adding only works if you're using pre-multiplied alpha. – Mark Ransom Sep 12 '14 at 14:23

How to programmatically fill in RGB noise in the transparent region of an image using Python?

1 Answers1