iDeLog3D: Sigma-Lognormal Analysis of 3DHuman Movements

Abstract

This paper proposes a 3D representation of human kinematics with the Kinematic Theory of Rapid Human Movements and its associated Sigma-Lognormal model. Based on the lognormality principle, a human movement is decomposed as a vector sum of temporally overlapped simple movements called strokes, described as two virtual target points linked by an arc of circumference and with the movement velocity having a lognormal shape. The paper extends the former 2D theory to the third dimension by linking the 3D virtual target points with planar circumferences covered with lognormal velocity profiles and reconstructing the 3D kinematics of the whole movement with temporally overlapping consecutive planes. Parameter optimization is accomplished jointly in the temporal and spatial domains. Moreover, the lognormal parameters used are numerically estimated, potentially providing a set of possible solutions that gain insights into the physical and biological meanings of the Sigma-Lognormal model parameters. We show that the 3D model, called iDeLog3D, achieves competitive results in analyzing the kinematics of multiple human movements recorded by various sensors at different sampling rates. The iDeLog3D is available to the scientific community following license agreements.