抄録
This paper describes a method for construction of a feature points-based shape model using visual ID tags and natural feature points in the camera images. The method involves the construction of a polyhedral model of the object by means of visual ID tags assigned to natural feature points of the object. The pose of the target object are estimated by the matching of the feature points in the captured camera image and those of the constructed feature-points model. The surface of the object model are projected to the camera image using the pose-estimation results, therefore, the estimation system can register the feature points to the object shape model. The paper focuses on the evaluation of the estimation of the pose of the object model. Experimental results show the feasibility of the method.
