論文
高速M推定型ガウシアンフィルタとガウシアンフィルタの特性を応用した外れ値の検出・除去と粗さ評価
キーワード:
outlier,
roughness profile,
mean line,
low-pass filter,
gaussian filter,
robust gaussian filter,
FMGF
2020 年 86 巻 10 号 p. 778-783
詳細
2020 年 86 巻 10 号 p. 778-783
If an outlier exists, an accurate roughness evaluation is not possible. The solution of this problem is insufficient only by calculating a mean line and roughness profile by robust filter. In addition to it, excluding of the outlier region is necessary. Fast M-estimation based Gaussian filter (FMGF) is the robust Gaussian filter using the fast M-estimation method that is developed so far by authors. The FMGF shows robustness only for the outliers such as spike noise, then the output agrees with the Gaussian filter output when no outlier exists. In this paper, the region of the outlier is extracted from the roughness profile by using the characteristic of FMGF and GF. And the roughness parameter is calculated from the data that excluded the spike noise, without needing manual data processing.
