Problem with adding Noise to a grayscale image with Python

Asked Jul 18 '21 at 17:56

Active Jul 18 '21 at 17:56

Viewed 335 times

I tried to add a gaussian noise to a grayscale image with something like:

noise = np.random.normal(0, sigma, img.shape(img))
img = img + noise

The problem is that i got a noisy image when i save it with io.imsave() but not with same one that i used in Matlab, there is some differance in contrast i think because of this statement :Lossy conversion from float64 to uint8. Range [-1713.9923310279846, 1488.4250569343567]. Convert image to uint8 prior to saving to suppress this warning.

So even when i tried to convert it to uint8 i get as a result, an image with white points or event a dark image with some white points. Any idea to solve this lossy conversion that didn't let me get the right noisy image as i want.

asked Jul 18 '21 at 17:56

salim terai

`img.shape(img)` does this even work? It should give an exception as `img.shape` is a `Tuple` – smttsp Jul 18 '21 at 18:03
Also, what is `sigma`? I would have followed something like this: https://stackoverflow.com/a/30609854/712452 – smttsp Jul 18 '21 at 18:06
No, it's np.shape(img). It was a mistake when i wrote this topic. – salim terai Jul 18 '21 at 18:13
sigma, is a value for intensity of the noise. More like standard deviation var – salim terai Jul 18 '21 at 18:14

Problem with adding Noise to a grayscale image with Python

0 Answers0