"... In PAGE 7: ... It can be observed that the algorithm is able to recover the correct motion and that rotation estimates tend to be more accurate than translation estimates. Table2 shows the error of the relative scales be- tween consecutive translations. We can see that the scale is estimated with an error below 7%.... ..."

"... In PAGE 6: ... The residual error matrix of Equation 6 was decomposed into its six independent components for analysis. Table2 gives the angular and translational residuals for fluo- roscopic guidance. Table 3 gives the angular and translational residuals for computer- assisted guidance.... ..."

"... In PAGE 6: ... Table 2 gives the angular and translational residuals for fluo- roscopic guidance. Table3 gives the angular and translational residuals for computer- assisted guidance. For both techniques, the mean and standard deviations of the residu- als is shown.... ..."

"... In PAGE 6: ...specific listing of these error types with example errors is given in Table1 (syntactic errors) and Table 2 (non-syntactic errors). Since Grammatifix is under constant development, an up-to-date version of its error types is available on the Internet ( Arppe et al 1999).... ..."

"... In PAGE 12: ... The only change that needs to be made is that the prior distribu- tion (2) needs to be updated given the information from the short-term measurement. We thus replace M and S2 in the above formulas by Mnew D M=S2 C yst log b = 2 st 1=S2 C 1= 2 st ; S2 new D 1 1=S2 C 1= 2 st ; (14) where yst is the logarithm of the short-term mea- surement, b is the correction factor derived from Table2 and st D log 1:8 . If the short-term mea- surement was not made in winter, then a seasonal correction factor will also apply; see, for example, Mose and Mushrush (1997) and Pinel et al.... In PAGE 12: ... 3. If a short-term measurement has been taken, up- date the prior distribution using (14) and the bias correction from Table2 (and possibly an addi- tional seasonal correction). 4.... ..."

"... In PAGE 7: ... The error reduction of the combined model compared to the letters-only model is substantial: for 1-Best, the error reduction is over BEBFB1; for 2- Best, 3-Best, and 4-Best it is even higher, reaching over BGBIB1 for 4-Best. As an example of the influence of pronuncia- tion modeling, in Table4 we list some misspelling- correct word pairs where the LTR model made an incorrect guess and the combined model CMB guessed accurately. 5 Conclusions and Future Work We have presented a method for using word pro- nunciation information to improve spelling correc- tion accuracy.... ..."

"... In PAGE 8: ... In Appendix 2 we show how this model can be written in state space form and estimated using the Kalman filter and maximum likelihood methods. Table2 reports the 9 See Smets (1998). This estimation methodology extends the work by Kuttner (1994) and Gerlach... In PAGE 17: ... First, following the literature discussed above we analyse the optimal Taylor rule if we take into account the estimated variance-covariance matrix of the parameter estimates as a measure of model uncertainty. The lower panel of Table2 presents the results.23 We basically confirm the results of Estrella and Mishkin (1998) and Rudebusch (1998) who show that parameter uncertainty only marginally reduces the efficient feedback parameters in the instrument rules.... ..."

"... In PAGE 13: ... The model is estimated separately for each sub- sample as well as for the entire sample. Table2 presents the parameter estimates for the whole sample as well as for all sub-samples.... ..."