LAUSR

In this paper, the frequency‐based decomposition approach (FBDA) for preprocessing of surface texture parameters calculation/assessment was proposed. Two types of textures were analysed: turned or grinded details. They were measured with interferometric method. For detection/minimization of high‐frequency measuring errors (HFME) the wavelets, Haar, Daubechies, Coiflet, biorthogonal or reverse‐biorthogonal approaches were proposed. Moreover, in some cases, the power spectrum density and/or autocorrelation function analysis provided an additional and/or valuable information about surface texture/features as well. It was assumed that application of various extraction methods (such as FBDA) can provide reasonable results for noise extraction in accordance to the commonly used algorithms, such as median filter, moving average (arithmetic mean) filter or other denoising procedures (e.g., Gaussian denoising filter) proposed in commercial software. The influence of proposed procedure on calculation (error minimization) of surface topography parameters (from ISO 25178 standard) was also presented. Some errors (especially those caused by false estimation of noise‐removal procedure) in surface topography parameter calculation were comprehensively studied with profile analysis; for areal HFME decomposition, the noise surface was defined. For some turned details, the HFME was observed in profiles analysis mostly. The biggest high frequencies from removed noise surface (created by noise extraction with FBDA) contained, the better results (of proposed approach) were obtained. It was also assumed that FBDA for preprocessing of both grinded and turned surfaces with various decomposition filters (in some cases the RBW) gave more ‘direct results’ that commonly used denoising approaches.