code for white gaussian noise for image

Hi, I have a Lena image with size 512X512 and I want to add white Gaussian noise with mean=0 and variance=10 to this image. do you have any code that do this for me? thanks in advance.

Answer 1

[Namwon Kim]

%% Code for White Gaussian Noise for Image

% noisy = (sqrt((Standard Deviation)^2)*randn(size(Lena_image))+mean + Lena_image

% Where (Standard Deviation)^2 is a variance, and

% [512, 512] = size(Lena_image)

Therefore,

load Lena % Input: Lena image
noisy = (sqrt(10)*randn(512,512))+0 + Lena;

Answer 2

[Walter Roberson]

https://www.mathworks.com/help/images/ref/imnoise.html

[nadia]

I need additive white Gaussian noise(AWGN), Do this function produce this type of noise for me?

[Walter Roberson]

https://www.mathworks.com/matlabcentral/answers/24282-image-processing-noise

[nadia]

can you please guide completely me . I want to add additive white Gaussian noise to Lena with variance 10. I am a beginner. do you have any code? thanks

[Image Analyst]

You just follow the directions. The only "trick/catch" is that the variance assumes the image is in the range 0-1 so you can either use im2double() or you can divide your variance by 255^2.

grayImage = imread('lena.jpg');
subplot(1, 2, 1);
imshow(grayImage);
title('Original Image', 'FontSize', 30);
noisyImage = imnoise(grayImage, 'gaussian', 0, 10/255^2);
subplot(1, 2, 2);
imshow(noisyImage);
title('Noisy Image', 'FontSize', 30);
diffImage = double(grayImage) - double(noisyImage);
variance = var(diffImage(:))  % Check that it's around 10

[nadia]

thank you very much for your answer. if my image has values between [0 1], so, the above code is correct for me or not?

[Image Analyst]

No. You'd use 10 instead of 10/255^2 because your max value is 1, not 255. You will have an extremely noisy image. So much so that you probably won't be able to see your underlying original image.

[nadia]

ok, if I want to change this variance (var=10) so that it suitable for double images, how can I change it? I have a watermark program and I want to attack it with AWGN but my images are double. So, if I want to have a noise with this variance but suitable for double images, how can I change the variance?

[Walter Roberson]

10/255^2 for that case.

[nadia]

are you sure? when I consider your solution with my double images(values are between [0 1]) the variance of difference of original image and noisy image is not about 10?

[Walter Roberson]

A variance of 10 is not "suitable or double images" (that are in the range 0 to 1). Especially not if you think of the range 0 to 1 as being upper and lower bounds on representation and do not permit (say) -7 to +7 to be stored there to give room for a clear variance of 10. If you clamp at 0 to 1 then you can never get a variance of 10.