DIGITAL FILM RESTORATION AND IMAGE ENHANCEMENT USING ARTIFICIAL INTELLIGENCE

Overview

Research project

Digital Film Restoration and Image Enhancement using Artificial Intelligence

Faculty

Engineering

Project head

Prof. Dr.-Ing. Wolfgang Ruppel

Prof. Dr.-Ing. Michael Voigt

Participants

M. Eng. Sebastian Marcus Edelmann

Duration

01.05.2022 – 30.04.2024

Project description

Digitizing historical film reels and preserving cultural heritage is a challenging social task. There is a lot of old film material that has not yet been restored to a state worth seeing due to the high costs and the large amount of time required for restoration. The goal of the project is an innovative solution for AI-based digital film restoration and image enhancement. This will automate sub-processes to simplify and accelerate restoration and technical enhancement.

Film restoration is about restoring the original image impression of old film recordings as accurately as possible. This involves reconstructing damaged image content and correcting faded and color-cast film sequences. In image enhancement, the technical characteristics of the film recordings are improved, possibly resulting in a different image impression. Image enhancement includes the removal of noise and blur as well as the increase of resolution, frame rate, dynamic range and color gamut. There is also the possibility of image stabilization of shaky material and colorization of black and white films.

Although both tasks have different requirements, they can still be solved using similar methods of artificial intelligence. For this purpose, established algorithms are adapted and further developed for use on old film material. The AI models are trained and optimized with scans of real film to achieve maximum convergence.