A RULE BASED APPROACH FOR IDENTIFYING A POSSIBLE 3D LINE DRAWING OBJECT FROM A SCANNED 2D IMAGE

摘要

Retrieving and identifying a 3D object from a database exists in several engineering and science applications. In this paper, the problem of extracting and identifying a possible 3D line drawing object from a scanned 2D image is considered. Several previous studies related to this problem have reported with the assumption that a scanned image is assumed to contain a 3D object projected onto a 2D image space. Our study does not retain this assumption. In addition, the edges and vertices of the scanned object image are not provided in advance. The proposed algorithm locates each line by applying Radon transform. Then, each vertex formed by three or more intersecting lines is identified. Finally, all identified vertices are systematically used to extract all the essential incident edges. The algorithm is tested with a set of selected 250 polyhedral 3D objects obtained from Varley's thesis. All objects can be correctly identified.