1A1-M02 複雑な背景画像下における視覚IDタグと画像特徴を用いた三次元物体モデルの生成 : モデル利用時における物体モデルの逐次更新(空間知)

抄録

This paper describe a method for construction of an object shape model and estimation of the three dimensional pose (position and orientation) of the object using the shape model based on a combination of visual ID tags and three-dimensional natural feature points. We construct a feature points-based shape model by capturing visual ID tags and feature points of the image with a monocular camera. To estimate the object pose, the method assigns the 3D feature points to the capture image using SIFT features. In this paper, we focus on update of the object model using the pose-estimation results and pose estimation of a marker-less object. Experimental results show feasibility of the proposed method.