The Demodulation Algorithms For Periodically Three-Dimensional Vibrating MEMS Based On Stroboscopic Phase-Shift Interferometry

With the quickening development of Micro-Electro-Mechanical Systems(MEMS)technology and the wide range of its applications,the demand for dynamic testing of the MEMS devices is increasing.Dynamic tests are extremely useful in the research,manufacturing,and quality control of the MEMS devices.They also have significant implications for improving the performances and reducing defect rates of the devices.To meet the demands of the dynamic test,we built a stroboscopic phaseshift interferometer to capture the transient image sequences of the periodically moving MEMS devices.To demodulate the motion parameters from the images,we developed a feature-point-matching(FPM)algorithm based on Speeded-Up Robust Features(SURF)for in-plane motion detection and a bi-phase unwrapping algorithm in the time and space domain for off-plane motion detection.We successfully achieved high-precision three-dimensional motion detection of the MEMS devices at a low cost.The main work of this paper is summarized as follows.1.The stroboscopic phase-shift interferometer was built,which mainly consisted of a stroboscopic drive control module,a phase-shift interference module,a lightemitting diode source,and an imaging system.By controlling the strobe signal and its phase delay,as well as the CCD signal,the system was able to visualize the MEMS devices at a frequency of 500 k Hz using ultrafast stroboscopic photography utilizing20 ns light pulses.It can acquire clear images of the in-plane and off-plane motion beyond the resolution limit of human eyes.In the phase-shift interference module,the piezoelectric actuator drove the nano-positioning stage to generate phase shifts,and the interferograms at different phase shifts were acquired by the imaging system.2.The SURF-based FPM algorithm was developed to demodulate the motion parameters of the MEMS devices with sub-pixel precision from the image sequences.The proposed methods consisted of three major steps: feature point detection,feature point matching,and mismatched point-pairs elimination using the Random Sample Consensus(RANSAC)model.The accuracy of the in-plane demodulation was 100 nm,and the processing rate was 1884 fps,which can realize real-time detection.3.A bi-phase unwrapping algorithm in the time and space domain was developed to demodulate the off-plane motion parameters.We obtained the phase distribution on the surface of the measuring device by extracting phase from the acquired interferograms with the five-step phase-shifting method,reducing noise with SUSAN and spin filtering,and unwrapping the phase both in the time and space domain.After converting the phase information to height,the off-plane motion parameters were finally calculated.