Feature-Based Characterisation of Turned Surface Topography with Suppression of High-Frequency Measurement Errors.

Errors that occur when surface topography is measured and analysed can be classified depending on the type of surface studied. Many types of surface topographies are considered when frequency-based errors are studied. However, turned surface topography is not comprehensively studied when data processing errors caused by false estimation (definition and suppression) of selected surface features (form or noise) are analysed. In the present work, the effects of the application of various methods (regular Gaussian regression, robust Gaussian regression, and spline and fast Fourier Transform filters) for the suppression of high-frequency measurement noise from the raw measured data of turned surface topography are presented and compared. The influence and usage of commonly used available commercial software, e.g., autocorrelation function, power spectral density, and texture direction, which function on the values of areal surface topography parameters from selected (ISO 25178) standards, are also introduced. Analysed surfaces were measured with a stylus or via non-contact (optical-white light interferometry) methods. It was found that the characterisation of surface topography, based on the analysis of selected features, can be crucial in reducing measurement and data analysis errors when various filters are applied. Moreover, the application of common functions can be advantageous when feature-based studies are proposed for both profile and areal data processing.