| 摘要 |
The precision of measurements of feature dimensions of objects using a particle beam device is improved by setting upper and lower signal reference levels and a signal offset level, performing a preliminary scan of the object using the particle beam device to produce a time-quantized and level-quantized signal indicative of a surface profile of the object, determining a percentage of time intervals during which the signal lies outside a range defined by the upper and lower signal reference levels, and processing the signal by adjusting at least one of the signal reference levels and the offset level such that the foregoing percentage becomes substantially equal to a predetermined percentage. In other words, the percentage of "outliers" in a profile, that is the number of pixels that exceed full scale in a video range, is limited to a predetermined percentage, for example one percent. Clipping distortion is therefore minimized. In another embodiment of the present invention, clipping distortion is minimized by performing profile synthesis using the median of multiple profiles instead of the arithmetic mean.
|
