Video object segmentation using graphs

Abstract

This paper presents an approach for video object segmentation. The main idea of our approach is to generate a planar, triangulated, and labeled graph that describes the scene, foreground objects and background. With the help of the Kanade-Lucas-Tomasi Tracker, corner points are tracked within a video sequence. Then the movement of the points adaptively generates a planar triangulation. The triangles are labeled as rigid, articulated, and separating depending on the variation of the length of their edges. © 2008 Springer-Verlag Berlin Heidelberg.