Improving Output Visualization of an Algorithm for the Automated Detection of the Perceived Origin of Movement

The perceived Origin of full-body human Movement (OoM), i.e., the part of the body that is perceived by an external observer as the joint from which movement originates, represents a relevant topic for movement analysis. Indeed, its automated detection is important to contribute to the automated analysis of full-body emotions and of non-verbal social signals, and has potential applications, among others, in dance and music teaching, cognitive and motor rehabilitation, sport, and entertainment. In this work, we further develop a recently proposed algorithm for the automated detection of the perceived OoM, by improving the visualization of its output. Specifically, the core of that algorithm relies on clustering a skeletal representation of the human body based on the values assumed by a movement-related feature on all its vertices, then finding those vertices that are at the boundary between any two resulting clusters. In the work, we improve the visualization of the clusters generated by that algorithm in successive frames, by “colouring” them by means of the resolution of a sequence of minimum cost bipartite matching subproblems. Finally, based on a real-world dataset, we show that the proposed modification of the algorithm provides, indeed, a better visualization of the clusters than its original version.