next up previous contents
Next: Generalization to 3D Shapes Up: Tests Previous: Robustness of the Method

   
Comparison with Reference Measurements

The optical gauge system was applied to a large set (100 pieces) of E-cores whose dimensions had been manually measured at the Polfer (Warsaw, Poland) ferrite core producing factory using the currently standard gauging technique. High degree of correlation between the manual and the computer-based, optical measurements was observed.

An example of an E-core is shown in figure 2. The dimensions measured are explained in figure 10. Here, we present the results of measurements for one of the dimension, namely, $A$.


  
Figure 10: The measured dimensions of an E-core.
\begin{figure} \begin{center} {\epsfig{figure=/users/mitya/illustr/crash/mgv96/ecore.eps,width=0.60\linewidth} } \end{center} \end{figure}

The comparison is represented by the three plots shown in figures 11, 12 and 13. Figure 11 is the distribution of the $A$ values obtained with the previous matching procedure [6]. Figure 12 was obtained by the presented HD-based procedure, while the last plot shows the reference manual measurements done by the core producer (Polfer). The plots testify that the three measurements are consistent, with the HD-based one being closer to the manual one in the shape of the distribution, which is single-peak.


  
Figure 11: Results of the original matching procedure.
\begin{figure} \begin{center} {\epsfig{figure=/users/mitya/illustr/squash/meeting4/sav_A1h_gl.ps,width=0.65\linewidth} } \end{center} \end{figure}


  
Figure 12: Results of the HD-based matching procedure.
\begin{figure} \begin{center} {\epsfig{figure=/users/mitya/illustr/squash/meeting4/sav_A1hd.ps,width=0.65\linewidth} } \end{center} \end{figure}


  
Figure 13: Results of manual measurements.
\begin{figure} \begin{center} {\epsfig{figure=/users/mitya/illustr/squash/meeting4/sav_A1pol.ps,width=0.65\linewidth} } \end{center} \end{figure}

next up previous contents
Next: Generalization to 3D Shapes Up: Tests Previous: Robustness of the Method
Dmitry Chetverikov
1998-11-16