Interactive Media Systems, TU Vienna

Video Object Segmentation Using Stereo-derived Depth Maps

By Danijela Markovic and Margrit Gelautz


Abstract currently not available.


D. Markovic, M. Gelautz: "Video Object Segmentation Using Stereo-derived Depth Maps"; in: "Vision in a Dynamic World", C. Beleznai, T. Schlögl (ed.); issued by: Austrian Association for Pattern Recognition (OEAGM/AAPR); Österreichische Computer Gesellschaft, Wien, 2003, ISBN: 3854031688, 197 - 204.

Additional Information

In this paper, we explore possibilities to improve existing video segmentation algorithms by utilizing a stereo-derived depth map as additional source of information. The goal is to make the segmentation results more robust in order to reduce the need for user interaction in typical video editing and compositing tasks. In tests with synthetic and real video scenes, we show that a combination of the original and stereo-derived edges in conjunction with active contour models (snakes) can improve the quality of the segmentation results.


Click into the text area and press Ctrl+A/Ctrl+C or ⌘+A/⌘+C to copy the BibTeX into your clipboard… or download the BibTeX.