Restoration of noisy radiographic images applied in Non Destructive Testing (NDT)

Restoring degraded images is a problem that is part of the digital image processing domain. When acquiring images, phenomena such as noise, blur and bad quality are always present. Our goal is to reduce noise in the case of radiographic images applied in Non Destructive Testing (NDT) to get closer to a more authentic image. In our simulation we considered a Gaussian noise and a real noise, we use methods based on minimization algorithms without constraints such as fixed step gradient, optimization algorithms with constraints such as the projected gradient algorithm and an algorithm that applies a regularization (Total Variation) based Rudin-Osher-Fatemi model (ROF) using the Chambolle projection to improve the quality of the results. At the end a comparative study of algorithms used.