This paper presents and discusses an effective approach to assessing human postural deviations based on curvature measurements. Multiscale curvature values are calculated from both parametric contours extracted from the silhouette image in sagittal plane and from the medial axis. The algorithms were applied to digital images of patients who have been submitted to Global Postural Reeducation (GPR) physiotherapy treatment. Features such as area, perimeter, center of mass, spreading, thinness degree and angulations are also obtained for similarity shape analysis between images taken before and after the GPR treatment. The medial axis is evaluated to investigate how it can be used to infer the spine alignment of patients with postural deviations prior to taking exams such as x-ray or tomography.
展开▼
