Multiple alignments of range maps by active stereo imaging and global marker framing

Three-dimensional shapes can be digitised by using active imaging techniques, which reconstruct entire objects by capturing multiple range maps from different viewpoints. Multiple-view reconstructions require the computation of translation and rotation parameters to transform each range map with reference to a global coordinate frame. In this paper, an automatic method has been developed to efficiently align 3D range maps acquired by an active stereo vision system. The methodology is based on referring range maps to a global frame of fiducial markers captured by the stereo system. The procedure includes a refinement of the marker frame in order to globally minimise the misalignment errors. The methodology optimises the overall accuracy of 3D reconstructions regardless of the scanning strategies, even processing large data sets. The proposed approach has been experienced and validated by measuring both nominal shapes and industrial models.