SUMMARY

See the on-line review and/or wait for hardcopy publication, soon.

This paper concerns itself with tracking hand motion. The main problem faced is with a large degree of freedom and it is addressed by seeking reasonable constraints in "a non-parametric feasible configuration space representation." A two-stage hierarchical simplex search has difficulties to address: in image feature extraction, with cost functions and convexity matters. Enter "multiple hypotheses" Monte Carlo simulation. Roots for all aspects of the approach and prior work by the authors are cited, leading to a lot of background information. The amount may be too much for an 'average' reader while it is our view on the other hand that more information would be needed to actually follow up on the work. The method is demonstrated for ten motions. Could be a good starting point for the paper, since it makes some preceeding information clearer.