Comparison of Different Image Recognition Methods for Characterization of Micro-gyroscopes

摘要

In order to have an efficient design methodology and achieve high-performance inertial instruments, it is important to analyze the dynamic characteristics of micromechanical vibratory gyroscopes. In this paper, a novel optical method based on a high-speed CMOS camera and different image recognition methods is developed to measure the translational motion of the sensitive elements. Experimental results show that digital image correlation (DIC) in the spatial domain is more accurate for determination of small displacements while quite time-consuming, and the mode of addition, subtraction or multiplication of spectrum images is much faster and provide more accurate results for determination of large displacements. Due to the edge features of images, the edge detection method based on wavelet is also proposed to calculate the displacements. This technique is quite fast and its precision is moderate. Accordingly, the combination of different image recognition methods will be more suitable for dynamic displacement measurements which require both high accuracy and high speed.