Advanced techniques for separation of roughness, waviness and form.

摘要

Filtering is the important process in characterizing engineering surface texture. After reviewing current filtering techniques, detail study for multiband filters that include bandpass filters and filter bank has been performed. Finally, multiresolution filtering based on wavelet transform is introduced and its potential application in surface metrology is investigated comprehensively in this study. The fundamental idea in wavelet is to analyze signal according to scale. Multiresolution filtering decomposes surface signals at different times and frequencies by different resolutions. The filtered signal can be combined to obtain the original profile using the perfect reconstruction ability of multiresolution filtering technique. By extensively investigating basic wavelet concepts, implementation approaches, and comparisons among the characteristics of available wavelets, Coiflet4 wavelet is selected in this study due to its good transmission characteristics and near linear phase response. The steps of multiresolution analysis in surface metrology involve decomposing surface profiles, separating surface features and reconstructing them functionally. The potential applications are surface feature extraction, functional evaluation, and process monitoring. This provides a new, powerful and flexible tool to reveal the multiscale features of surfaces.