Study of the blur, ringing and aliasing artifacts in numerical imaging : application to restoration
Gwendoline Blanchet · HAL (Le Centre pour la Communication Scientifique Directe) · 2006
This thesis deais with the problems related to the formation of numerical images. Sampling, which is a necessary stage for the formation of a discrete image from a continuous one, may introduce some artifacts that degrade the image quality. These artifacts are called blur, ringing and aliasing. The main motivation in this thesis was the study of these three artifacts. In the first part, we recall the image formation process and we define these artifacts. In the second part, we propose a new measure of both the ringing and blur artifacts associated to a low pass filtering prior to sampling. The third part is dedicated to the automatic detection of these artifacts in images. In the fourth part, the automatic detection is tested on two real restoration applications: the blind deconvolution and the denoising.