Temporal Video Segmentation Using Global Motion Estimation And Discrete Curve Evolution (MA-P4)

Author(s) : Siripong Treetasanatavorn (Siemens AG CT IC 2, Germany) Joerg Heuer (Siemens AG CT IC 2, Germany) Uwe Rauschenbach (Siemens AG CT IC 2, Germany) André Kaup (University of Erlangen-Nuremberg, Chair of Multimedia Communications and Signal Processing, Germany) Klaus Illgner (Siemens AG CT IC 2, Germany)

Abstract : Temporal video segmentation is considered a fundamental yet important step for several high-level video analysis tasks. In contrast to most traditional approaches, the paper contributes to temporal video subshot segmentation that complements the conventional shot-scene-sequence hierarchical paradigm. Global motion is assumed carrying implicit semantics applicable in temporally segmenting a run of camera shot into meaningful subshots. The method treats the problem as a discrete form identification and exploits the discrete curve evolution technique in analysing such subshots of pronounced global motion characters. Experimental results based on several standard test sequences acknowledge uses of the method.

Menu