Study of sub-pixel image processing algorithms for MEMS in-plane vibration measurements by stroboscopic microscopy

Several optical methods have been developed for the measurement of in-plane vibration of microscopic objects. However most of them needs a scattering surface or a specific surface structuring. A low cost method which has not these limitations is optical stroboscopic microscopy combined with image processing by optical flow techniques. Previous works have shown that a nanometric sensitivity can be obtained. In this paper, we investigated several subpixel image processing methods for in-plane vibration measurements of MEMS by this technique. Emphasis was put on whole displacement field measurements and on fast algorithins able to process a large sequence of images without the need of a multi-resolution approach to get local vibration amplitudes. It is notably shown that use of spatiotemporal regularity between images is an efficient way to reduce noise and that a resolution in the 0.01-0.033 pixel range can be achieved. Results are applied to in-plane vibration local measurements in two perpendicular directions at video rate as well as to full-field mapping of in-plane vibration mode of electrostatically actuated MEMS devices in SOI technology.