Full 3D Rotation Estimation in Scanning Electron Microscope

International audienceEstimation of 3D object position is a crucial stepfor a variety of robotics and computer vision applicationsincluding 3D reconstruction and object manipulation. Whenworking in microscale, new types of visual sensors are used suchas Scanning Electron Microscope (SEM). Nowadays, microandnanomanipulation tasks, namely components assembly,are performed in teleoperated mode in most of the cases.Measuring object position and orientation is a crucial steptowards automatic object handling. Current methods of poseestimation in SEM allow recovering full object movement usingits computer-aided design (CAD) model. If the model is notknown, most methods allow to estimate only in-plane translationsand rotation around camera optical axis. In the literature,SEM is considered as a camera with parallel projection or anaffine camera, which means image invariance to z-movementand bas-relief ambiguity. In this paper, authors address theproblem of measuring full 3D rotation of the unknown scenefor uncalibrated SEM without additional sensors. Rotations areestimated from image triplets by solving a spherical trianglefrom fundamental matrices only, without need of intrinsiccalibration, allowing to avoid parallel projection ambiguities.The presented results, obtained in simulation and on real data,allow validating the proposed scheme