Characterizing the surface waviness of steel sheet: reducing the assessment length by robust filtering

In this work we investigate the possible reduction of the measuring length Lt for the assessment of the waviness of steel sheet. Different parameters and filter techniques are compared regarding their behaviour as a function of Lt. Application of a standard Gauss filter leads to inaccurate results, even in case of sufficiently large values of Lt, by the introduction of artificial waviness due to an incorrect deviation of the reference centre line. For smaller Lt, there is the additional disadvantage of the loss of data when using such a filter. On the other hand, promising results obtained with a Robust Gauss filter. In particular, the waviness parameter based on this filter is extremely stable upon reducing Lt.