Model-based recognition of 3D objects using intersecting lines

Exploiting geometric features, such as points, straight or curved lines and corners, plays an important role in object recognition. In this paper, we present a model-based recognition of 3D objects using intersecting lines. We concentrate on using perpendicular line pairs to test recognition of a parallelepiped model and represent the visible face of the object. From 2D images and point clouds, first, 3D line segments are extracted, and then intersecting lines are selected from them. By estimating the coverage ratio, we find the most accurate matching between detected perpendicular line pairs and the model database. Finally, the position and the pose of the object are determined. The experimental results show the performance of the proposed algorithm.