Person Identification from Full-Body Movement Using String Grammar Fuzzy-Possibilistic C-Medians

Biometrics is a method to identify a person. However, there are several biometric techniques including face recognition, palm recognition, iris recognition, fingerprint recognition, and so on. There is a new approach in biometrics, i.e., identification using full-body movement. In this paper, we introduce a full-body movement in human identification using three Kinects. In particular, we utilize the string grammar fuzzy-possibilistic C-medians (sgFPCMed) to group string sequences from skeleton frames into pose string, then group the pose string sequences of each person into multi-group to create multi-prototypes for each person. The K-nearest neighbor is used to identify the person in the test process on 27 subjects. The system yields 73.33% correct classification on the best validation set of four-fold cross validation.