Robust human motion modeling strategy and its application on driver identification and gait recognition

A robust and compact human motion model is desirable in many security applications from public facilities to personal devices. Shape features are extracted from the perspective of computer vision in most researches. However, most of them are application-dependent. In order to explore more dynamical features of human motion and to make the human model adaptable to the varying environments, a new robust modeling strategy which utilizes pre-knowledge from feature space analysis is proposed. The contribution of this work is not only the integration of necessary statistical tools on data analysis, but also the proposal of a special modeling strategy: extending robust estimation methods against outliers in input space to robust modeling strategy against undesirable features in feature space. The proposed model has a circular topology consisted of 2 pairs of correlated states and a constrained Expectation Conditional-Maximization (ECM) algorithm is proposed under the model´s unique structure.