A new procedure for characterizing textured surfaces with a deterministic pattern of valley features

In recent years there has been the development of a high number of manufacturing methods for creating textured surfaces which often present deterministic patterns of valley features. Unfortunately, suitable methodologies for characterizing them are lacking. Existing standards cannot in fact properly characterize such surfaces, providing at times unreasonable values. In this paper, a new procedure for characterizing such surfaces is proposed, relying on advanced filtering and feature recognition and separation. Existing advanced filtering methods do not always eliminate all distortions, therefore some modifications are investigated. In particular the robust Gaussian regression filter has been modified providing an envelope first-guess in order to always fit the mean line through the plateau region. Starting from a filtered and aligned profile, the feature thresholds recognition and separation is therefore made possible. In particular the plateau and valley regions can be detected, separated and independently analyzed according to their function. The example of a multifunctional profile is presented. The profile is analyzed using the new procedure, which demonstrates outputting a correctly aligned roughness profile and permitting comprehensive feature analyses.