Image Inpainting | Inverse Problem | L1 and TV Priors | Proximal algorithm | FISTA | python
Image Inpainting: Solving Inverse Problem with L1 and (anisotropic) TV Priors with Proximal point algorithm FISTA with python
For L1 prior the Wavelet operator was used to transform to a sparse representation
Original cameraman graysacle image was corrupted by randomly discarding 40% of the pixels (with mask), the rest of the pixels were used in the proximal point optimization algorithm FISTA
Different values of the regularization hyperparameter were used to test the reconstructed image quality with PSNR
As can be seen, TV prior obtains better quality of the output image, when compared with original input image
#imageprocessing #imageprocessingpython #python #algorithm #computervision #machinelearning #sparse