A Generative Model for Textured Motion Analysis and Synthesis

• Activity Number: 137
• Type: Topic Contributed
• Date/Time: Monday, August 12, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Computing*
• Author(s): Yizhou Wang*+ and Song Chun Zhu
• Affiliation(s): Ohio State University and Ohio State University
• Keywords: Motion Analysis ; Textured Motion ; hidden Markov models ; generative models ; Markov chain Monte Carlo

Abstract

Natural scenes contain rich stochastic motion patterns, which are characterized by the movement of a large number of small elements, such as falling snow, flock of flying birds, firework, and waterfall. We call these motion patterns "textured motion." The analysis and synthesis of such motion patterns are important for a variety of applications in both vision and graphics and stimulate growing interest of the two communities. By building up a generative model combined with statistical models and algorithms for the textured motion, we try to learn realistic motion patterns from real data, separate the motion dynamics with photometric and geometric styles, and thus achieve good controllability and interpretation of motion (by movemes' trajectories, sources, sinks, force map, shadings, etc.).