POSE AND SUB-POSE CLUSTERING-BASED IDENTIFICATION OF INDIVIDUALS

摘要

This theme discloses the system and method for the identification of individual. This method includes obtaining each individual of static and one frame frame of dynamic feature vector skeleton data to carry out step activity and arbitrary graphic pattern and in random walk; It creates, each individual, the frame in the first predetermined number of clusters dynamic feature carrier; Creation, each individual, each cluster of the associated frame dynamic feature vector of each cluster of interior submanifold of the second predetermined quantity; And determination, for it is each individually, propose data set based on gait feature calculate center dynamic feature vector correlation frame each sub-cluster, the feature vector of averaged static frame and each cluster, the individual for identification.