Consider a simplified example:


noise = Table[
    NormalDistribution[1, 0.2]] Sin[Sqrt[x^2 + y^2]]^8, {x, -16, 
   16 - 16/512, 32/512}, {y, -16, 16 - 16/512, 32/512}];

original image

img1 = ExampleData[{"TestImage", "Mandrill"}];

image with noise

img2 = ImageAdd[img1, Image[0.2 noise]]


enter image description here

Is it possible to remove the noise and reconstruct the original image just from the noised image, assuming that the circles are all centered at the center of the image?

  • $\begingroup$ I would make a histogram of pixel values versus radius, which will surely reveal the periodic noise structure, then balance and project back to the original pixel space. $\endgroup$ – David G. Stork Oct 19 '17 at 0:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.