Temporal super resolution using variational methods

Sune Høgild Keller, Francois Bernard Lauze, Mads Nielsen

Research output: Chapter in Book/Report/Conference proceedingBook chapterResearchpeer-review

Abstract

Temporal super resolution (TSR) is the ability to convert video from one frame rate to another and is as such a key functionality in modern video processing systems. A higher frame rate than what is recorded is desired for high frame rate displays, for super slow-motion, and for video/film format conversion (where also lower frame rates than recorded is sometimes required). We discuss and detail the requirements imposed by the human visual system (HVS) on TSR algorithms, of which the need for (apparent) fluid motion, also known as the phi-effect, is the principal one. This problem is typically observed when watching video on large and bright displays where the motion of high contrast edges often seem jerky and unnatural. A novel motion compensated (MC) TSR algorithm using variational methods for both optical flow calculation and the actual new frame interpolation is presented. The flow and intensities are calculated simultaneously in a multiresolution setting. A frame doubling version of our algorithm is implemented and in testing it, we focus on making the motion of high contrast edges to seem smooth and thus reestablish the illusion of motion pictures.

Original languageEnglish
Title of host publicationHigh-quality visual experience : creation, processing and interactivity of high-resolution and high-dimensional video signals
EditorsMarta Mrak, Mislav Grgic, Murat Kunt
Number of pages22
VolumePart III
PublisherSpringer
Publication date2010
Pages275-296
Chapter11
ISBN (Print)978-3-642-12801-1
ISBN (Electronic)978-3-642-12802-8
DOIs
Publication statusPublished - 2010
SeriesSignals and Communication Technology
ISSN1860-4862

Cite this