A Statistical Similarity Measure for Aggregate （总体的统计相似性度量）.pdfVIP

A Statistical Similarity Measure for Aggregate Crowd Dynamics Stephen J. Guy⇤ Jur van den Berg Wenxi Liu, Rynson Lau Ming C. Lin, Dinesh Manocha University of Minnesota University of Utah City University of Hong Kong UNC-Chapel Hill Figure 1: A comparison between a rendering of real-world crowd data (a), and stills from three different simulation algorithms applied to the same scenario (b-d). Our entropy metric is used to measure the similarity of simulation algorithm to real-world data. A small value of the metric, as in (d), indicates a better match to the data. Differences between the simulations are highlighted with circles. Abstract 1 Introduction We present an information-theoretic method to measure the similar- Visual simulation of aggregates systems, including human crowds, ity between a given set of observed, real-world data and visual sim- animal herds, and insect swarms is a growing area of interest in ulation technique for aggregate crowd motions of a complex system computer graphics, with applications in diverse areas such as social consisting of many individual agents. This metric uses a two-step sciences, swarm intelligence, and city planning. For applications process to quantify a simulator’s ability to reproduce the collective in entertainment, providing artists and animators with high-level behaviors of the whole system, as observed in the recorded real- control while maintaining visual plausibility of motion is often suf- world data. First, Bayesian inference is used to estimate the sim- ficient. However, for many other training and planning applica- ulation states which best correspond to the