Rotationally- and scale-invariant pattern recognition of tribological surfaces

A new method, called a pattern recognition by partition iterated function system (PR-PIFS), has recently been developed and successfully applied to classify wear particles. The method is based on the fact that the surface topography of 3-D objects can be described by a small set of mathematical rules. Since the PR-PIFS method is fully automated, it eliminates the need for specialists and reliance on human visual inspection techniques. Also, the method is rotationally- and scale-invariant and no vector of various surface parameters is needed. The application of this method to the classification of artificially generated fractal and tribological surfaces (e.g. worn surfaces) is presented in this paper. For this purpose, the PR-PIFS method has been specially modified to suit tribological surface data. The accuracy of this method in the classification of surfaces rotated by various angles and scaled to different sizes was assessed.