Federica Sciacchitano: During acquisition and transmission, images are often blurred and corrupted by noise. One of the fundamental tasks of image processing is to reconstruct the original image from a degraded version. Additive white Gaussian noise has been extensively studied, since it produces simple and tractable mathematical models.
However, in many real applications, the noise is much more complicated, and cannot be well simulated by additive white Gaussian noise. For example, it can be signal dependent, multiplicative, or even mixed.
This PhD project focuses on the development of image reconstruction models and algorithms under non-Gaussian noise, such as multiplicative noise, impulse noise, Poisson noise, Cauchy noise, etc. The new methods will reconstruct high quality images with less noise and more details, which will enhance image processing operations, such as edge detection, segmentation, etc., for the benefit of a wide range of applications in modern society. Moreover, since the non-Gaussian noise often appears in tomographic reconstructions, this project is strongly connected with the ERC project “High-Definition Tomography”.
Effective start/end date 01/09/2013 → 26/10/2016
Published as PhD report: Image reconstruction under non-Gaussian noise
Supervisors: Per Christian Hansen (co-supervisor), Yiqiu Dong (main supervisor)
Section for Scientific Computing