Estimation of surface roughness parameters on a base of machined surface image / Anna Zawada-Tomkiewicz.

Dane z autopsji.

Prediction of a machined surface parameters is an important factor in developing unmanned cutting centers. A new noncontact technique to assess the deferences in the quality of the product is of great need. The metod suggested in this paper is based on a machine vision system for surface roughness estimation. There have been wavelet methods introduced in the application for machined surface image analysis. Digital image of a machined surface was described with the use of one-dimensional Digital Wevelet Transform (Coiflet). The statistical description of particular wavelet decomposition levels made possible the estimation of surface roughness parameters with the use of recurrent neural networks.