FMDistance: A Fast and Effective Distance Function for Motion Capture Data

Given several motion capture sequences, of similar (but not identical) length, what is a good distance function? We want to find similar sequences, to spot outliers, to create cl usters, and to visualize the (large) set of motion capture sequences at our disposal. We propose a set of new features fo r motion capture sequences. We experiment with numerous variations (112 feature-sets in total, using vari ations of weights, logarithms, dimensionality reduction) , and we show that the appropriate combination leads to near-p erfect classification on a database of 226 actions with twelve different categories, and it enables visualiza tion of the whole database as well as outlier detection.