The analysis of human movements, gathering quantitative information about the musculoskeletal system during the execution of a motor task, plays a fundamental role for the understanding of the system physiopathology. Current methods for capturing accurate 3D human motion require the attachment of devices like markers, fixtures or sensors, that may influence the execution of natural movements, require time for subject preparation, and must occur in laboratory environment. Markerless, video-based, techniques are emerging as a solution to overcome these problems, but their application in clinical practice and biomechanical research is still limited because no good solution yet exists to accurately estimate bone movements from standard video. In particular, observed deformations of soft tissues in video data may alter predictions of bone location. An unexplored step in the direction of building a reliable video-based technique for markerless bone motion estimation is to design a model to accurately infer the movement of the underlying bone from the external structure. The design of such model is the main objective of this research and will be enabled by simultaneously captured bi-plane X-ray and video sequences. Being able to predict the bone movements from external measurements can open up non-invasive analysis in natural environments that can significantly improve our understanding of articulated body kinematics. The outgoing phase will take place at Brown University (USA), where a group of researchers have recently set up a state-of-the-art 3D X-ray imaging system for visualizing the complex bones and joints motion during natural activities. The Brown team also has developed state of the art methods of markerless human shape and motion analysis. The return phase will take place at the Movement Analysis Laboratory, Istituti Ortopedici Rizzoli (Italy), where the outcome of the research will be applied clinically for motion analysis in healthy subjects and patients.