Novel auto-regressive measurement of diamond tool wear in ultra-precision raster milling

Abstract

In this paper, a new auto-regressive algorithm is proposed to reconstruct 3D topographic surface for diamond tool wear, using its in-process image. First, based on digital image processing technique, tool wear lands are separated from a tool wear image captured by a CCD camera under a 100X optical system; Second, a traverse search method of arc translation is put forward to eliminate feigned wear lands, and a least square polynomial method is adopted to fit inner- and outercontours of the wear lands, self-adaptively filtering noises and connecting the discontinuous wear lands; Finally, the autoregressive calibration method is developed to reconstruct its 3D topographic surface. The wear land is extracted selfadaptively, and the wear area, maximal wear width, average wear width and worn volume can be determined automatically by the algorithm. The reconstructed 3D topography of the tool wear land can be identified, based on the tool wear image captured by SEM. The result indicates that the method is capable of reconstructing 3D topography of the tool wear land and provides a possibility for in-process 3D-wear measurement in ultra-precision raster milling (UPRM) and the algorithm reliability is validated finally. And the influence of tool wear on surface roughness is discussed.