Localization and tracking of a 3-D object based on multi-view image acquisition

This paper presents a 3-D object localization and tracking technique based on the CAD model and multi-view image captures of the object. From the given projected 2-D pose model in the image, the matching lines between the model contour and the object´s edge feature are used for nonlinear 3-D pose computation. The object location information in the real world is then identified. The method presented in this work has been validated on several experiments with various test objects. The results demonstrate that the proposed approach is robust to the partial occlusion and provides accurate positioning of the real object.