为提高光学干涉测量的精度, 研究了用来消除干涉条纹图噪声的滤波算法。讨论了不同波面和不同干涉仪噪声下不同滤波方法的滤波效果。考虑干涉仪噪声对条纹图的影响与检测中的不同波面情况, 基于泽尼克多项式原理用matlab仿真法建立了离焦、像散、彗差、球差4种不同的干涉条纹图, 计算了滤波后不同波面面形的PV值和GRMS值差别的大小。结果表明: 同种噪声与滤波器情况下, 在离焦、像散、彗差、球差4种面型中, 滤波后球差波面的PV差值最大, 高达0.11λ, 与在乘性噪声下其它波面的最高PV值0.08λ相比, 滤波效果较差; 滤波后像散波面的PV差值最小, 除乘性噪声中值滤波情况外, PV差值均在0.05λ以下, 最低达0.01λ, 滤波效果较好。此外, 彗差波面滤波情况要略优于离焦波面。滤波器中, 低通滤波稳定性高, PV差值局限于0.01λ~0.04λ, 滤波效果相对好; 而空域滤波后波面的PV差值最高达0.09λ, 但噪声针对性更好。

To improve the measuring accuracy of optical interferometry, the filtering methods to eliminate the noise in interference fringe patterns were researched. The results of different filtering methods were discussed in different wavefronts and different noises of the interferometer. By taking the effects of interferometric noise on fringe patterns and different wavefronts into account, four kinds of fringe patterns for defocus, astigmatism, coma, spherical aberration were established based on zernike polynomials by matlab simulation. The differences of PV and GRMS values before and after filtering were calculated for different wavefronts. The results on four kinds of wavefronts for defocus, astigmatism, coma, spherical aberration show that under the same kind of noise and filtering, the PV difference of the spherical aberration is largest, up to 0.11λ, which shows a poor filtering effect as compared with the 0.08λ from other wavefront differences under the multiplicative noise. Furthermore, the PV difference of the astigmatism wavefront is smallest, and it is down to 0.05λ excepting median filtering in the multiplicative noise, showing a better filter effect. Moreover, the result of coma wavefront filtering is slightly better than that of the defocus wavefront filtering. For the filter methods, low-pass filters offer high-stability and the differences of the PV is from 0.01λ to 0.04λ, showing a better filtering effect.