Three-dimensional computerised video analysis is becoming increasingly important in the clinical assessment of disorders of movement, especially gait. Unfortunately, the accuracy of these systems is limited by the accuracy of placement, and relative motion of, skin markers over bony landmarks.
Recent developments in image-recognition now allow the outlines of moving objects to be tracked automatically. The proposed project would apply and adapt these techniques to clinical motion analysis, to provide marker-less tracking. This would not only be more convenient for the operator, but has the potential to improve accuracy and reliability.
A combination of deformable models, B-spline curve representation and control theory will be used to track the three-dimensional motion of the lower-limb segments of a walking subject from two synchronous video-images. Results will be compared with those obtained from conventional motion analysis using markers.