Artefact-reduced kinematics measurement using a geometric finger model with mixture-prior particle filtering

Abstract

It is challenging to measure the finger's kinematics of underlying bones in vivo. This paper presents a new method of finger kinematics measurement, using a geometric finger model and several markers deliberately stuck on skin surface. Using a multiple-view camera system, the optimal motion parameters of finger model were estimated using the proposed mixture-prior particle filtering. This prior, consisting of model and marker information, avoids generating improper particles for achieving near real-time performance. This method was validated using a planar fluoroscopy system that worked simultaneously with photographic system. Ten male subjects with asymptomatic hands were investigated in experiments. The results showed that the kinematic parameters could be estimated more accurately by the proposed method than by using only markers. There was 20–40% reduction in skin artefacts achieved for finger flexion/extension. Thus, this profile system can be developed as a tool of reliable kinematics measurement with good applicability for hand rehabilitation.

Keywords::

• finger kinematics
• particle filter
• motion analysis
• skin artefact

Acknowledgements

The authors would like to thank National Science Council of Taiwan, ROC, for their support under contract NSC 95-2221-E-006-239-MY2. We are also indebted to the members of Department of Radiology, National Cheung Kung University Hospital and the members of the Visual System Lab, Department of CSIE, NCKU, for their valuable discussions and their participation in clinical experiments during the course of this study. Furthermore, we would like to thank the unknown reviewers and the editor for their comments.

Notes

^1. Email: [email protected].

^2. Ratio of skin–artifact reduction = (marker-based error–proposed error)/(marker-based error).