By Kin-Hong Wong and Michael Ming-Yuen Chang
Department of Computer
Science and Engineering
The Chinese University of Hong Kong
Last updated on 29th June, 2004.
Key words: 3D object reconstruction,
bundle adjustment, 3D object scanner, structure from motion
Most of the vision-based scanner systems available today use a turntable to rotate an object so that multiple views of the object can be captured by a stationary camera. The ultimate accuracy of the 3D reconstruction of the object frequently lies on an optimization technique known as bundle adjustment. We note, however, that for these systems, the object¡¦s motion is constrained to rotate around a fixed axis. By incorporating this constraint into the traditional bundle adjustment method, we show that the reconstruction accuracy can be substantially improved. This ¡§constrained¡¨ bundle adjustment method will be useful in the design of low-cost, yet accurate, 3D scanners for use in manufacturing and entertainment industry.
Demo Video: cba_demo1.mpg