Gait kinematics features, which concerns its geometry, can be represented using its spatial and temporal characteristics; namely moments of silhouette and angular displacements of limb. Limb angular displacements data are manual labeled hip angle, knee angle, and ankle angle data and silhouette moments data are simple centralized moments of binary image data. This paper presents an analysis on spatial and temporal features of gait kinematics. A standard approach of using PCA (principal component analysis) and CA (canonical analysis) algorithms are used for the analysis. To ensure an invariant analysis due to differences in walking speed, a cycle extraction procedure, which consists of interpolation and resampling, is performed beforehand. Results from this paper can suggest that angular displacements data can be a better feature representation in comparison to the simple silhouette moments data.
Download Full PDF Version (Non-Commercial Use)