Accurate 3D shape and displacement measurement using a scanning electron microscope.

摘要

With the current development of nano-technology, there exists an increasing demand for three-dimensional shape and deformation measurements at this reduced-length scale in the field of materials research. Images acquired by Scanning Electron Microscope (SEM) systems coupled with analysis by Digital Image Correlation (DIC) is an interesting combination for development of a high magnification measurement system. However, a SEM is designed for visualization, not for metrological studies, and the application of DIC to the micro- or nanoscale with such a system faces the challenges of calibrating the imaging system and correcting the spatially-varying and time-varying distortions in order to obtain accurate measurements. Moreover, the SEM provides only a single sensor and recovering 3D information is not possible with the classical stereo-vision approach. But the specimen being mounted on the mobile SEM stage, images can be acquired from multiple viewpoints and 3D reconstruction is possible using the principle of videogrammetry for recovering the unknown rigid-body motions undergone by the specimen.; The dissertation emphasizes the new calibration methodology that has been developed because it is a major contribution for the accuracy of 3D shape and deformation measurements at reduced-length scale. It proves that, unlike previous works, image drift and distortion must be taken into account if accurate measurements are to be made with such a system. Necessary background and required theoretical knowledge for the 3D shape measurement using videogrammetry and for in-plane and out-of-plane deformation measurement are presented in details as well. In order to validate our work and demonstrate in particular the obtained measurement accuracy, experimental results resulting from different applications are presented throughout the different chapters. At last, a software gathering different computer vision applications has been developed.