Study on Vision-Based Multidirectional Posture and Motion Analysis System Development

Abstract

In this study, a multidirectional posture and motion evaluation tool system using a camera-based vision system was developed. By installing a camera on the top, back, left, and front of the system, kinematical information of the participant was collected and analyzed on the sagittal, coronal, and transverse planes. A dedicated color light-emitting diode (LED) marker was developed for increasing recognition rate of marker. Based on LabVIEW, images collected from four cameras were printed on one screen, and the detection angle of each marker was calculated using the cosine second law. To evaluate the performance of this system, a motion analysis experiment was performed on 20 office workers in Jeonju area. The reliability was measured using POFOMO, Dartfish and the results were compared. The comparison results indicate that the concurrent validities of the two results for the knee angle, left shoulder, and right shoulder were highly correlated as [Formula: see text], 0.988, and 0.991, respectively. Because the POFOMO image analysis tool in the present system reduces the analysis time and involves a simpler measurement procedure than that of Dartfish, this system can be used as an effective device for measuring the posture alignment of the human body.