The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Modeling of objects using planar facets in noisy range images
Author:	Phansalkar, Girish
View Online:	njit-etd1997-084 (vi, 71 pages ~ 3.6 MB pdf)
Department:	Department of Mechanical Engineering
Degree:	Master of Science
Program:	Mechanical Engineering
Document Type:	Thesis
Advisory Committee:	Dave, Rajesh N. (Committee chair) Caudill, Reggie J. (Committee member) Ji, Zhiming (Committee member)
Date:	1997-01
Keywords:	Computer graphics. Fuzzy algorithms. Pattern perception.
Availability:	Unrestricted
Abstract:	Products designed and manufactured before the advent of Computer Aided Design (CAD) and Computer Aided Manufacturing (CAM) technology have not been documented electronically. To avoid the laborious procedure of redesigning the parts, a reverse engineering approach can be adopted. This approach involves, taking a picture of the object and constructing a solid model from the image data. Range image is a three dimensional image of an object or a scene. This image can be obtained from special cameras, called range image cameras, or can be constructed from the Coordinate Measuring Machine's (CMM) output data. Adaptive Fuzzy c-Elliptotype (AFC) clustering algorithm is used to identify the planar facets in a range image. A modified version of AFC algorithm can handle noisy range images. Unknown number of planar facets can be identified using the Agglomerative clustering approach. The object is reconstructed using segmented image data. The equations of the edge are obtained from the plane intersections. An edge validity criterion is developed to validate the existence of an edge. Vertices are the two extreme points on the edge. A Boundary representation of the object is developed. The information about this object is then passed to a CAD software using Initial Graphics Exchange Specification (IGES).