Influence of Different Filtration Methods Application on a Filtered Surface Profile and Roughness Parameters

Tatiana Miller,Aneta Łętocha,Krzysztof Gajda
DOI: https://doi.org/10.4028/www.scientific.net/kem.637.57
2015-02-01
Key Engineering Materials
Abstract:The measurements were performed on the surfaces made of different materials and typical of diversified character. Glass roughness standards with sinusoidal profile, approximately sinusoidal profile and metallic comparative standards after lathing and grinding were object of research. Analysis was performed including the surface and profile evaluation. Statistical analysis was conducted. Measurement sections and other filter parameters were selected in accordance with standards. Measurements were carried out with stylus tip contact method – using TOPO 01P device designed by The Institute of Advanced Manufacturing Technology, that uses diamond tip inductive sensor. Tip sensor radius is equal to 2 μm. The results of measurements were filtered by: Gaussian filter, Robust Gaussian regression Filter, Spline, Spline Wavelet, Morphological Filter. Gaussian Filter uses linear system based on Fourier wavelengths. Robust Gauss Regression Filter is similar to Gaussian Filter, but it is insensitive on the specified phenomena in input signal. Spline Filter is based on linear polynomial combination. Wavelet Filter decomposes profile on constant shape elements, but on different scales. Morphological Filter operates on the principle of filtered profile plotting using circular disc or horizontal line segment with a specified (respectively) radius or length. Selection of suitable filtration method is essential and one of the most important things to obtain reliable measurement results evaluation. Not all filters are suitable for each type of surface. Filter algorithms differ from each other and this influences in a greater or lesser degree on the roughness profile and hence on roughness parameters and waviness parameters related to it.
What problem does this paper attempt to address?