Human Identification Based on Gait Motion Capture Data

The authors present results of the research aiming at human identification based on gait motion capture data. Second-order tensor objects were chosen as the appropriate representation of data. High-dimensional tensor samples were reduced by means of the multilinear principal component analysis (MPCA). For the purpose of classification the following methods from the WEKA library were used: k Nearest Neighbors (kNN), Naive Bayes, Multilayer Perceptron, and Radial Basis Function Network. The maximum value of the correct classification rate (CCR) equal to 95.71% was achieved for the classifier based on the multilayer perceptron.