中文摘要：

气阀配合面表面粗糙度决定着燃气表的机械与物理性能。为了加快表面粗糙度评定速度，根据高斯权函数的对称性，推导了高斯滤波器的快速卷积算法；采用光谱共焦传感器扫描工件表面轮廓的方式，比较了线性高斯滤波、回归高斯滤波、稳健高斯滤波在气阀配合面表面粗糙度提取过程中的优缺点。结果表明：新型稳健高斯滤波不仅能解决边界畸变问题，还能消除深谷或高峰等异常信号的影响，提高了表面粗糙度评定精度。

英文摘要：

Surface roughness of diaphragm gas valve fitting face determines its mechanical and physical properties. In order to speed up the evaluation of surface roughness, the fast convolution algorithm of Gaussian filter was derived based on the symmetry of Gauss function. With the means of spectral confocal sensor scanning surface profile of workpiece, the advantages and disadvantages of linear Gaussian filtering, regression Gaussian filtering and robust Gaussian filtering were compared. The results indicate that new robust Gaussian filter can not only solve the problem of boundary distortion, but also eliminate the effects of valley or peak abnormal signals. The evaluation accuracy of surface roughness has been improved.