Extraction of the dominant substructure of fracture surfaces in 3D

When fracture surfaces are rough and irregular, it is not possible to extract their dominant objects automatically. A method of locally controlled filtering (LCF) is proposed for achieving this aim, inherently targeted at the dominant substructure of input data. The degree of reduction of surface shapes is corrected by one numerical parameter, which can be selected as free or estimated from the input data. A characteristic for comparing outputs, called the parameter of reduction, is defined. An application on a 3D sample of the impact fracture surface of steel P92 is presented and compared with other solutions of filtration using Fourier and wavelet transforms and trap-ezoidal filtration windows. Although the solution using wavelets is more effective, both trans-formations lead to a greater content of undesirable smaller objects in the output in comparison with LCF, under the same value of the parameter of reduction. Several applications of fully automatic LCF on various fractures of steels P92 and AISI 304L are presented.

Automated summary

A locally controlled filtering (LCF) method is proposed to automatically extract dominant structures from rough and irregular fracture surfaces. The reduction of surface shapes can be adjusted by a numerical parameter, and a feature called the parameter of reduction is defined for comparing outputs. An application on a 3D sample of steel P92 fracture surface is presented and compared with other filtration solutions using Fourier and wavelet transforms and trap-ezoidal filtration windows. Despite wavelets being more effective, both transformations result in a higher presence of undesired smaller objects in the output compared to LCF under the same reduction parameter value. Various applications of fully automatic LCF on fractures of steels P92 and AISI 304L are demonstrated.