To Do: image denoising in wavelet domain
Data
Images you want (you can use Barbara provided in the Fourier section)
To do
- Simulate various inpainting problem as follow
- Generate a random binary matrix A of the same size of the image, with a parameter $p$ controling the bernoulli law
- Add some white gaussian noise (at various level)
- Generate the direct problem $y = Ax + b$ (where $x$ is the original image)
- Estimate $x$ using the sparse approach (reminder: an image is sparse in the wavelet domain)
- Discuss the results obtained by changing:
- the sparse respresentation (various wavelet orhtogonal transform and translation invariant wavelets)
- the thresholding rules (soft, hard, empirical Wiener)
- the choice of the $\lambda$ parameter
- Discussion whould be made with respect to the value of $p$ and the level of noise