Modeling shape, motion, and flexion of non-rigid 3D objects in a sequence of images

摘要: A method models a non-rigid three-dimensional object directly from sequence of images. shape the is represented as matrix 3D points, and basis possible deformations displacements points. The matrices points forming model object. Evidence for an optical flow determined image intensities in local region near each point. evidence factored into rotation, translation, deformation coefficients to track video.