Fig. 2

Outline of the registration method. The initial position of the bone model on the first frame is defined manually, and virtually projected images are generated for the bone models. Edge-enhanced fluoroscopic images and virtually projected images, F and V, are obtained, and the positions and orientations of the bones are calculated based on the similarity measure (Equation 1). Template-matching is used to obtain the initial guess of bone positions and orientations on the next frame, and the registration proceeds in a consecutive manner