Estimating the 3D orientation of a microgripper by processing the focus data from the images delivered by a videomicroscope

Visual servoing is a key technique to automate microassembly tasks since videomicroscope is widely use to recovery informations about the scene. The parts as well as the gripper should be tracked in the images which do not deliver a complete focused view of the objects because of the weakness of the depth-of-field. The paper investigates the problem of computing the 3D orientation of a rigid body by analysing the focus data extracted from the images of a videomicroscope. The solution is based on the shape-from-focus approach. But the modelling stage is simplified, instead of performing a 3D modelling, a 1D modelling is achieved which enables the estimation of the 3D orientation of the object. The concept is applied with success to the tips (70 mum width and 1500 mum length) of a gripper imaged by a 10times videomicroscope.