Optic Flow Estimation for Human Spine Movements Using a Motion Capture System

Human interaction technology based on motion capture (MoCap) systems is a vital tool for human kinematics analysis, with applications in clinical settings, animations, and video games. We introduce a new method for analyzing and estimating dorsal spine movement using a MoCap system. The captured data by the MoCap system are processed and analyzed to estimate the motion kinematics of three primary regions; the shoulders, spine, and hips. This work contributes a non-invasive and anatomically guided framework that enables region-specific analysis of spinal motion which could be used as a clinical alternative to invasive measurement techniques. The hierarchy of our model consists of five main levels; motion capture system settings, marker data acquisition, generation of the dorsal spine skeleton, optic flow calculation in 3D space, and motion protocol analysis and evaluation. Points of interest (POIs) are identified at ten locations on the three selected regions; the acromion of the scapula, the vertebral column, and the greater trochanter of the femurs. The optic flow of these POIs is calculated based on their spatiotemporal positions in 3D space that facilitates the analysis of the articulated spine motion. We rigorously tested our model with various movement protocols including flexion, lateral-flexion, and twisting movements. The results demonstrated an exceptional performance with a minimal error of $(2.09^{\circ })$ compared with ground truth data. These findings highlight the accuracy of our model in estimating optic flow and tracking the motion of human spine.