Perceptual Categorization of Movement Blends

Abstract

Techniques of computer animation, allied with traditional visual psychophysics, provide potentially useful tools for the exploration of the perception of human actions. In this ongoing research we examine the use of motion blends to study action recognition. By blending from one action to another we obtain tuning functions for action recognition. It is hoped that elucidation of the processes and properties that underlie the recognition of action will be of use for the testing and design of blending algorithms.