Motion graphics, description and control

Abstract

Perhaps the single most important reason to investigate methods of motion graphics, description and control is that these modalities verify that a representation is adequate to describe motion phenomena. Whether designed for human or computer vision, a motion representation often only serves as a target for information extracted to accomplish a particular task, such as pattern recognition, classification, change detection, or shape description. Often this motion representation is selected by convenience and need only distinguish between phenomena of interest, possibly incorporating rather rudimentary notions of direction, velocity, and shape. Such representations may exhibit notable failings when they are asked to reconstruct what they purport to represent. Research in computer graphics, robotics, and expert systems is now confronting these representation questions directly by requiring the understanding and implementation of the semantics underlying a motion representation system. The representations can no longer be just instruments of convenience; rather they must be rich enough to simulate a reasonable slice of reality.