METHOD OF ESTIMATING SURFACE PROFILE PARAMETERS BASED ON PROBABILISTIC-STATISTICAL CLASSIFICATION OF THE PROFILOGRAM SPECTRUM Russian patent published in 2019 - IPC G01B21/30 G06N7/00 

FIELD: information technology.

SUBSTANCE: invention relates to evaluation of surface structure based on profile information. Technical result is achieved due to the fact that at the training stage for each profile type a plurality of samples is prepared, the profilograms are removed from the surfaces of the samples, the profilogram of the surface is presented in form of a signal, obtaining a signal spectrum, using sequences of spectral coefficients as class standards corresponding to the type of surface, probabilistic model is trained on a set of samples with clustering of the profilogram spectrum, obtaining statistical estimates of parameters of the model of the class, classifying the profilogram spectrum by selecting a model with the maximum likelihood value, wherein at the learning stage, the probabilistic model is adapted to specific features of the profilogram spectrum by using an arbitrary metric clustering algorithm, classes are formed by grouping sample profiles by one or a set of parameters of surface quality or surface properties, and evaluation of profile parameters is performed as a procedure for classifying the spectrum of a profilogram.

EFFECT: possibility of taking into account complete information on surface profile by analyzing spectral coefficients of profilogram.

1 cl, 4 dwg