Here are some simulations I did for image denoising
This model is proposed in L. Rudin, S.
Osher and E. Fatemi, "Non Linear Total Variation Based Noise
Removal Algorithms", Physica D, Vol. 60, pp. 259-268, 1992
Given the noisy image f, one tries to find the clean
image u as the minimizer of the energy
The associated Euler-Lagrange equation can be derived
as
The parameter need
to be optimized. Small introduces more
smoothing in u while large keep more
noise in u.
I show two examples here to demonstrate the algorithm.
First I add Gaussian noise to the images and then denoise them. (left:
original image , center: noisy image, right: denoised image)