Abstract
Variations in human gaits are manifest in the timing of the many combined motions in the gait. In periodic systems, such as a gait, timing reduces to phase. Therefore, in order to capture the important information in the timing patterns in a gait, one must consider phase. Gaits vary for several reasons, including different builds, moods of individuals, fatigue and injury. We investigate the relationship between the model-free shape-of-motion phase analysis and a subjective description of gait, such as a normal gait versus a tired gait or a shuffle, by analyzing several gait image sequences that differ subjectively. A simple model based on a phasor representation of gait motion relates the pendulum-like motion of limbs to shape-of-motion features. Our ultimate goal is to develop a gait feature space that can be partitioned according to subjective perception of gait. Gait features that vary with subjective changes in gait lead in this direction.