AUTOMATIC VIDEO OBJECT SEGMENTATION VIA VORONOI ORDERING AND SURFACE OPTIMIZATION I-Jong Lin S.Y. Kung Dept. of Electrical Engineering Princeton University Abstract zyxwvutsr - In this paper, we combine two theoretical concepts, Voronoi Ordering and Surface Optimization Formulation, to form a sys- tem for automatic Video Object Segmentation in support of the MPEG-4 standard. Voronoi Ordering is zyxwvu a means of projecting object shape onto the image space; the Surface Optimization Formulation provides fitness measure for object boundaries in the video sequence. By simultane- ously maintaining an invariant Voronoi Ordering and maximising the surface optimization metric, we can automatically extract Video Object Planes. This paper outlines the theoretical aspects of Voronoi Ordering and the formulation of Video Object Segmentation as a surface opti- mization problem. Results from a direct implementation of this theory are shown for three MPEG-4 test sequences (container, coastguard and hallmonitor). INTRODUCTION MPEG-4 ha est,ablisIied itself as the standard syntax for the first major step in content-based processing (e.g. cut and paste functionality, syut,liesis, query, et,c.). The niajor enabling technology for this standxd is a video object segmentation; given the amount of video, both archived and acquired daily, t,his segmentation process must be automatic. This paper advances a new concept of Voronoi Ordering and applies it wit,hin the frauiework of surface optimization systcm. hi our previous work zyxwvu [a], we formulated video object segmentation as a surface optimization problem. In our current work, we Lol- ster onr surface optinlization segmentation scheme by searnlessly integrating the concept of shape through the use of Voronoi Ordering. VORONOI ORDERING First, we begin with a formal definition of Voronoi Ordering. Definition 1 zyxwvutsrqp Given, zyxwvuts (L simple zyxwvutsrq closed contour c zyx within an image space I un11 a starting point 30 zyxwvut on C, we cun define 11 Voronoi Ordering EO) fvr- lJ, SbmIJk ClOSed COll~Oll~ c; &(;E I,:6) = miri(arg,cinin(Il:J- :Z(s)II)) (1) 0-7803-5610-1/99/$10.00 0 1999 265