| Abstract: | Conventional surface characterization techniques involving random process analysis are limited in characterizing multi-scale surface features relevant to manufacturing processes and functions. This paper introduces a novel technique for multi-scale characterization of engineering surfaces by applying wavelet transform. The main advantages of wavelet transform over other existing signal processing techniques are its space-frequency localization and multi-scale view of the components of a signal. Utilizing these properties of wavelet transform, we can effectively apply multi-channel filter banks to the surface data and link the manufacturing and functional aspects of a surface with its multi-scale features. Surfaces produced by typical manufacturing processes are analyzed using wavelet transform, and the usefulness of wavelet transform in the multi-scale analysis of engineering surfaces is demonstrated. |
