A first-order estimate of the data was found in order to determine the tilt of the sample.
Result of subtracting 1st order fitted curve from data. This should remove residual tilt. However, the drop-off at the right shows that there was a combination of tilt, rotational, and translational misalignment.
Given that removing first-order residuals of the data still left something to be desired, a higher-order fit was put onto the date. The goal was to remove other buried misalignment.
Removing a higher order curve attempts to eliminate the combination of the translation and tilt. However, it there was information missing from the measurement to be able to ensure that this is an accurate portrayal of the residual measurement.
gLike
Compare Misaligned Noisy Measurement to Design
Given the challenge of analyzing measurement data of an object floating in space, buried under noise and misalignment (i.e. tilt, translation, height, etc).
The measurement was meant to be compared to design values to monitor part degradation over time.
Simple analysis techniques were used to try to compensate for tilt, translational, and rotational misalignment.
A second iteration with a higher order analysis improved results but reduced robustness and reliability.