MODEL-BASED INSPECTION OF MACHINED SURFAC ES

Gray level images contain a lot of surface information completing the measurement with a conventional stylus gauge in a useful way. A signal model for groove textures in gray level images of machined surfaces i s proposed. The model has a parsimonious set of physically interpretab le parameters, which can be used as a specific feature set for objecti vely characterizing the essential geometric and statistic properties o f groove textures. Estimation algorithms for some of the parameters ar e given exemplarily. The performance of the model is demonstrated with synthetic textures. Finally, two applications are discussed: the dete ction of unwanted states in end milling processes and the automatic qu ality assessment of honed cylinder bores.