Title :
An integration framework for structure from motion estimation
Author :
Yang, L. ; Sahli, S.H.
Author_Institution :
Vrije Univ., Brussels, Belgium
fDate :
6/22/1905 12:00:00 AM
Abstract :
A framework for 3D structure/motion analysis is proposed based on the integration of different visual modules. The 2D motion module, which is characterized by normal flow of line segments, together with the feature based approach, are efficiently integrated with line drawing interpretations under our framework. Line drawing interpretations are introduced for image sequence analysis, and provide effective 3D structure information. An incremental scheme is developed to estimate 3D structure (such as line orientation and surface normal) from a sequence of images containing polyhedral objects. Experiments demonstrate the feasibility and advantages of the proposed integration framework for 3D structure estimation
Keywords :
image representation; image sequences; motion estimation; 2D motion module; 3D structure information; feature based approach; image sequence; incremental scheme; integration framework; line drawing; line orientation; line segments; motion analysis; motion estimation; polyhedral objects; structure; surface normal; visual modules; Image motion analysis; Image reconstruction; Image segmentation; Image sequence analysis; Integrated optics; Layout; Motion analysis; Motion estimation; Motion measurement; Shape;
Conference_Titel :
Signal Processing Proceedings, 2000. WCCC-ICSP 2000. 5th International Conference on
Print_ISBN :
0-7803-5747-7
DOI :
10.1109/ICOSP.2000.891676