Automatic film restoration: Metadata driven detection and removal of scratch and dirt

Motion picture films are often affected by scratch and dirt objects that reduce their quality. Manual restoration of digitized films is a time consuming and expensive business. We therefore present an automatic, three-step restoration process that generates editable metadata for quality control and reporting. For this, scratch and dirt detection and removal algorithms have been developed that do not require manual parameter adjustment. This solution is especially suited for restoration of large film archives with limited temporal and financial effort, but can also be applied to individual cases that require excellent quality.