Modern Plug-and-Play Image Priors for RAW Image Enhancement

Modern mobile phones have become the dominant photography device in recent years. However, due to form factor constraints both the camera sensor and lens have to remain compact, which in turn has a negative effect on the resulting image quality such as decreased resolution and noise. Image restoration attempts to recover a clean latent image from degraded input image(s) in a computational manner. To yield a natural image, one must impose additional constraints on the resulting optimization problem. In this project we explore some modern techniques for imposing such constraints, such as diffusion models and implicit image representation. We hope that this will not only result in an improved image quality but an easy plug and play tool to improve quality for any future imaging devices and methods. This is of major interest to Samsung as one of the leaders in the mobile phone industry.

Faculty Supervisor:

Igor Gilitschenski

Student:

Partner:

Samsung Electronics Canada

Discipline:

Computer science

Sector:

Technology; Information and Communications Technology; New and Digital Media

University:

University of Toronto

Program: