Research On3D Surface Profile Measurement Technology Of MEMS Based On White-Light Interference Principle

The micro electro mechanical system (MEMS) measurement’s main purpose is to providedata feedback for scientific research and product development, its one important aspect is thethree-dimensional (3D) surface profile measurement. Along with the development of MEMS,it is put forward higher requirements to measurement technology. Compared with traditional3D surface profile measurement technology, white light interference measurement technologyhas some characteristic such as large measuring range, non-contact, high sensitivity, nodamage, high precision and so on, has been widely used in the measurement of MEMS andother precision devices. At present, the research mainly focuses on improvement of measuringspeed, accuracy and other aspects, which is the mainly theme of this paper.Firstly, by analyzing the typical3D surface profile measurement methods and combiningthe most MEMS devices’ characteristics like steps, grooves and other discontinuous surface,the white light interference measurement technology was used. Elaborated the principle,equipment and measuring process of white light interference measurement technology.Then, combining with MEMS white light interference images, the mean filtering, medianfiltering, adaptive Wiener filtering, Butterworth low-pass filter, wavelet filter and other imagedenoising methods were discussed and compared. And then the evaluation standards of imagequality after filtering were analyzed. The center of gravity method, envelope curve fittingmethod, phase shifting method and space frequency domain algorithm were analyzed andcompared from the theoretical perspective. This paper proposed a method to extracting thewhite light interference signal envelope based on local peak point interpolation and presentedthe calculation process. The advantage of using cubic spline interpolation in the peak pointinterpolation method was discussed.Finally, the results of filtering methods comparison showed that, before subsequentprocessing of white light interference image, using median filtering method for preprocessing,could reduce the influence of noise on the measurement result. The peak point interpolationmethod, Fourier transform filtering method, Hilbert transform and wavelet transform methodwere compared from some aspects like measurement speed and accuracy, anti-noise abilityand scanning increment and so on through the Matlab software simulation. Through image preprocessing of median filtering method, white light interference signal processing of peakpoint interpolation method and combining with the single pixel size, the3D surface profilesof devices were obtained.