Abstract

Abstract: Matting aims to accurately extract a foreground object out of a photograph or video and is an important operation in many image/video editing applications. For instance, once an object has been extracted successfully from its background by appropriate matting techniques, it may be inserted into another scene. I will give an introduction to matting and discuss recent work in this area that was developed throughout my PhD studies in co-operation with Microsoft Research. More specifically, I will discuss different models for matting and present a ground truth dataset that can be used for a quantitative comparison of matting results.

Biography: Christoph Rhemann is a PhD student at Vienna University of Technology. He is currently working together with Margrit Gelautz, Carsten Rother (Microsoft Research) and Michael Bleyer in the field of computer vision with a focus on image segmentation and matting. His work was published at several international computer vision conferences including CVPR and BMVC .