Add noise to image-datastore using "Randn" command on Matlab

I have used command Randn, but still give me the same accuraccy result even after changing the noise magnitude. Why accuracy still same? how image can be affected by noise

used commands :

A=imread('human_6_77_.png');

K = imcrop(A,[116 50 675 534]);

b = 20*randn(size(K));

I=uint8(double(K)+b);

imshow(I)

[Sai Sri Pathuri]

Here are the images I tried with your code, i see a change in both images. There is more distortion for image with noise magnitude of 100. Can you attach your images? What did you mean by accuray? Is it the same as amount of distortion or something else?

% Original Image
figure
A=imread('coins.png');
K = imcrop(A,[116   50  675  534]);
I=uint8(double(K));
imshow(I)

% Image with Noise magnitude = 20
figure;
A=imread('coins.png');
K = imcrop(A,[116   50  675  534]);
b = 20*randn(size(K));
I=uint8(double(K)+b);
imshow(I)

% Image with Noise magnitude = 100
figure;
A=imread('coins.png');
K = imcrop(A,[116   50  675  534]);
b = 100*randn(size(K));
I=uint8(double(K)+b);
imshow(I)

[Mohamed Elbeialy]

As changing magnitude, the noise will increase on image. However, when trying to train this image on matlab "deep learning", it keeps givening me the same training accuracy evern with different noise magnitudes. why? what idology is running ?