Abstract:

To overcome the end effect generated by the convolution method of traditional Gaussian filter in obtaining surface midline, a new spline filter is constructed by variance difference to realize Gaussian filter. This algorithm, called spline realization of Gaussian filter, preserves both Gaussian filter and spline properties. This means that this filter not only accords with the transmission characteristics of ISO, but also restrains the end distortion on data sequence. The amplitude deviation between this algorithm and Gaussian filter falls the interval (-1.5%, +3.75%), which is smaller than the interval (-5.0%, +5.0%) of the recommended algorithm by ISO11562. In addition, using this kind of Gaussian algorithm, the computation of 11200 measured points takes only 100ms on current computer, which demonstrates the efficiency of the proposed algorithm.

Key words: metrology, profile filter, Gaussian filter, spline filter