抄録
The paper reports a RT Component for estimation 3D pose of a target object and design for usability by considering robot function. The RT Component extracts SIFT features from input images, and detect a target object by matching the SIFT features of the reference image to the SIFT features of the image. Then object pose is estimated based on correspondences of the features. Experimental results show the performance of the RT Component for several target objects. The RT component can detect a target object robustly in complex backgrounds of human daily life space.
