Behavior Segmentation Method Based on String Representation of Human Motion Capture Data

Abstract

The invention discloses a behavior segmentation method based on human body motion capture data character string representation. The method comprises steps of S1, taking human body motion capture data as multiple high dimension discrete data points, and calculating each Euclidean distance between data points; S2, performing clustering through a cluster method based on the local density and the relative distance of each data point to obtain a category to which the data point belongs, and using different characters for representing different categories; S3, reordering the characters according to time sequences of data corresponding to the characters to obtain character strings, and merging the same characters adjacent in time sequence of the character strings to form character groups, wherein each character group forms a behavior string; and S4, segmenting a whole behavior formed by the human body motion capture data according to the behavior strings, and extracting the motion period of each single behavior after segmentation. The technical scheme of the invention is good in accuracy, and excellent in applicability, validity and non-supervision.

Description

technical field [0001] The invention relates to the processing of human body motion capture data in computer animation. More specifically, it relates to an action segmentation method based on string representations of human motion capture data. Background technique [0002] Computer animation is the product of the combination of computer graphics and art. With the rapid development of computer graphics technology and computer software and hardware technology, computer animation has been widely used in many fields such as film and television special effects, 3D games, commercial advertisements, and computer simulation. [0003] In recent years, with the continuous development of hardware technology and the reduction of cost, motion capture systems have gradually become popular, and the optical-based three-dimensional human body motion capture (Motion Capture) method has developed into an important means of human motion information acquisition, with a large scale. Human body ...

AI Technical Summary

Problems solved by technology

However, this method cannot determine the same behavior contained in the sequence to be segmented, and there are many inconveniences in practical applications.

Zhou et al. from Carnegie Mellon University (F.Zhou, F.Torre and J.K.Hodgins..Hierarchical aligned clusteranalysis for temporal clustering of human motion..IEEE Transactions on Pattern Analysis and Machine Intelligence..2013, page: 582- 596.) The behavior segmentation based on human motion capture data is realized through the method of hierarchical alignment cluster analysis (Hierarchical Aligned Cluster Analysis, referred to as HACA). This method converts the behavior segmentation problem into an energy minimization problem, and uses the dynamic programming algorithm to realize Behavior segmentation, this method has a high segmentation accuracy, but this method requires the user to determine in advance the number of behaviors in the sequence and the number of clusters during time series reduction

Images