Perpustakaan Digital ITB

Abstract: This paper deals with development of object model control points extraction method from a monocular image. These control points are resulted from edge detection of straight line using linear regression method. This method is not influenced by the thickness of line segment, so that an enough accurate result can be achieved. The control points are then extracted to become 3D object model parameter. A knowledge structure for artificial intelligent algorithm is also proposed. The structure is based on connectivity laws among vertices. It is shown from several experiments that the proposed method give a good result in reconstructing the 3D object model consisting of orthogonal straight lines.