Automated Model Acquisition Using Volumes of Occlusion

Two primary requirements of any system that relies on active vision to perform modeling from observation are that it be able to use previously acquired data to drive the sensing process and that it can iteratively incorporate newly sensed data. This paper discusses an approach to automating CAD model acquisition by allowing the system to keep track of what parts of the sensed object have yet to be imaged. This is achieved by explicitly representing the volume of occlusion as well as its surfaces of the object obtained from any one sensing operation. Models built from distinct views of the object, and which include their volume of occlusion, are merged using set operations. The resulting composite model consists of the visible surfaces from each model along with the intersection of the volumes of occlusion and contains the necessary information for planning the next viewpoint.

[1]  Ramakant Nevatia,et al.  Description and Recognition of Curved Objects , 1977, Artif. Intell..

[2]  Henry Fuchs,et al.  On visible surface generation by a priori tree structures , 1980, SIGGRAPH '80.

[3]  M. Potmesil.  Generating three-dimensional surface models of solid objects from multiple projections , 1982 .

[4]  Gérard G. Medioni,et al.  Object modeling by registration of multiple range images , 1991, Proceedings. 1991 IEEE International Conference on Robotics and Automation.

[5]  Roger Y. Tsai,et al.  Automated sensor planning for robotic vision tasks , 1991, Proceedings. 1991 IEEE International Conference on Robotics and Automation.

[6]  Dmitry B. Goldgof,et al.  Building a B-rep from a segmented range image , 1994, Proceedings of 1994 IEEE 2nd CAD-Based Vision Workshop.

[7]  Glenn H. Tarbox,et al.  IVIS: An Integrated Volumetric Inspection System , 1995, Comput. Vis. Image Underst..

[8]  Bruce F. Naylor,et al.  Set operations on polyhedra using binary space partitioning trees , 1987, SIGGRAPH.

[9]  Roger Y. Tsai,et al.  Analytical characterization of the feature detectability constraints of resolution, focus, and field-of-view for vision sensor planning , 1994 .

[10]  Frank P. Ferrie,et al.  From uncertainty to visual exploration , 1990, [1990] Proceedings Third International Conference on Computer Vision.

[11]  Konstantinos A. Tarabanis,et al.  Computing Occlusion-Free Viewpoints , 1996, IEEE Trans. Pattern Anal. Mach. Intell..

[12]  R. Bajcsy,et al.  Three dimensional object representation revisited , 1987 .

[13]  Michael Werman,et al.  Active vision: 3D from an image sequence , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.

[14]  Harry Shum,et al.  Virtual reality modeling from a sequence of range images , 1994, Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS'94).

[15]  Paul J. Besl,et al.  Surfaces in Range Image Understanding , 1988, Springer Series in Perception Engineering.

[16]  John Amanatides,et al.  Merging BSP trees yields polyhedral set operations , 1990, SIGGRAPH.

[17]  R. Bajcsy,et al.  Shape recovery and segmentation with deformable part models , 1987 .

[18]  C. Ian Connolly,et al.  The determination of next best views , 1985, Proceedings. 1985 IEEE International Conference on Robotics and Automation.

[19]  Ramesh C. Jain,et al.  Invariant surface characteristics for 3D object recognition in range images , 1985, Comput. Vis. Graph. Image Process..

[20]  Anil K. Jain,et al.  Three-Dimensional Object Recognition Systems , 1993 .

[21]  Bahram Parvin,et al.  B-rep from unregistered multiple range images , 1992, Proceedings 1992 IEEE International Conference on Robotics and Automation.

[22]  Ruzena Bajcsy,et al.  How to Decide From the First View Where to Look Next , 1990 .

[23]  C A Steward,et al.  The state of the industry. , 1997, Clinical laboratory management review : official publication of the Clinical Laboratory Management Association.

[24]  B. Naylor.  A priori based techniques for determining visibility priority for 3-d scenes , 1981 .

[25]  Chien-Huei Chen,et al.  3d-poly: a robot vision system for recognizing objects in occluded environments , 1988 .