A topology-based approach for feature extraction from 3D motion gesture

In this study, we propose a topology-based approach for feature extraction from 3D motion gesture. First, we briefly review the topological invariants and their real-time parallel computing methods previously proposed. Next, we explain a topology-based touch sensing utilizing the Euler characteristicwhich is a fundamental topological invariant. Furthermore, we show that by expanding its framework the Euler characteristiccan be extracted from the trajectory of motion gesture as a feature in a real-time parallel and distributed manner. Finally, based on the results, we consider potential applications of our approach for human-machine interface.