블로그 이미지
Leeway is... the freedom that someone has to take the action they want to or to change their plans.
maetel

Notice

Recent Post

Recent Comment

Recent Trackback

Archive

calendar

1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31
  • total
  • today
  • yesterday

Category

2010. 9. 18. 11:58 Footmarks

4th Open Lab


September 14, 2010 | Written by admin

마침내 오픈랩이 다시 돌아왔습니다. 벌써 4번째 행사네요.
꼭 방문하셔서 제니텀의 최근 작업들을 경험하면서 즐거운시간 가지시길 바랍니다.

Zenitum’s Open Lab is back! Our Open Lab 4 will showcase our latest work in the field of augmented reality, including 3D reconstruction, and various techniques for recognizing and tracking images.


전시내용:


1. 영상기반의 모바일 증강현실 트래킹 엔진 & GPS기반 모바일 증강현실 트래킹 엔진

http://youtu.be/OvLTOWoze0A
http://youtu.be/YcgebgYeU5M
http://youtu.be/ibWnY9ZXKzk
http://youtu.be/7jUaxlS52tU
http://youtu.be/O-myIJboPn0


2. 4Cast: Full 3D 재구성 시스템

http://youtu.be/LByly6rlZMg
http://youtu.be/577gv_xeWPU
http://youtu.be/xL8YSgdQEXM

- 원하시는 분은 자신의 Full 3D 재구성 모델을 만들어 드립니다.


3. Media Art Project: iWall

- 3D 질감을 표현하는 대형 액티브 미디어 월과 iPhone과의 만남

- 기존(작년의 프로토타입)의 Active Media Wall 프로젝트 동영상은

http://youtu.be/wLlAfTa2lVg

posted by maetel
2010. 7. 29. 17:08 Computer Vision
Seitz, S. M., Curless, B., Diebel, J., Scharstein, D., and Szeliski, R. 2006. A Comparison and Evaluation of Multi-View Stereo Reconstruction Algorithms. In Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1 (June 17 - 22, 2006). CVPR. IEEE Computer Society, Washington, DC, 519-528. DOI= http://dx.doi.org/10.1109/CVPR.2006.19


S. Seitz et al. Multi-view stereo evaluation web page
http://vision.middlebury.edu/mview/
posted by maetel
Easy–to–Use Calibration of Multiple–Camera Setups
Ferenc Kahlesz, Cornelius Lilge & Reinhard Klein
University of Bonn, Institute of Computer Science II, Computer Graphics Group
http://biecoll.ub.uni-bielefeld.de
Proceedings of the ICVS(International Conference on Computer Vision Systems)  Workshop on Camera Calibration Methods for Computer Vision Systems - CCMVS2007 Published in 2007 by Applied Computer Science Group, Bielefeld University, Germany


posted by maetel
2010. 7. 20. 16:10 Computer Vision
Kai Ide & Steffen Siering & Thomas Sikora, "Automating Multi-Camera Self-Calibration", IEEE Workshop on Applications of Computer Vision, 2009

posted by maetel
2010. 7. 20. 15:41 Computer Vision
Kurillo, G., Li, Z., and Bajcsy, R. 2009. Framework for hierarchical calibration of multi-camera systems for teleimmersion. In Proceedings of the 2nd international Conference on Immersive Telecommunications (Berkeley, California, May 27 - 29, 2009). International Conference on Immersive Telecommunications. ICST (Institute for Computer Sciences Social-Informatics and Telecommunications Engineering), ICST, Brussels, Belgium, 1-6.

posted by maetel
2010. 7. 20. 15:17 Computer Vision
Ivo Ihrke & Lukas Ahrenberg & Marcus Magnor, “External camera calibration for synchronized multi-video systems”,
Journal of WSCG(the 12th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision), 2004                    
posted by maetel
2010. 7. 18. 17:05 Computer Vision
Accurate Camera Calibration from Multi-View Stereo and Bundle Adjustment
Yasutaka Furukawa & Jean Ponce, CVPR 2008

Yasutaka Furukawa
http://www.cs.washington.edu/homes/furukawa/

Jean Ponce
http://www.di.ens.fr/~ponce/
http://www-cvr.ai.uiuc.edu/ponce_grp/

DxO Labs (렌즈 왜곡 교정 소프트웨어)
http://www.dxo.com/

PMVS (Patch-based Multi-view Stereo Software) version 2
http://grail.cs.washington.edu/software/pmvs/

CMVS (Clustering Views for Multi-view Stereo)
http://grail.cs.washington.edu/software/cmvs/

sba : A Generic Sparse Bundle Adjustment C/C++ Package Based on the Levenberg-Marquardt Algorithm (Manolis Lourakis)
http://www.ics.forth.gr/~lourakis/sba/


image-based modeling

multi-view stereovision (MVS)

> two main approach camera calibration problem
1) chart-based calibration (CBC)
2) structure from motion (SFM) + auto-calibration + bundle adjustment (BA)

* selection of feature correspondences (SFC)
eg. RANSAC


Standard BA algorithms optimize both the scene point and camera parameters by minimizing the sum of squared reprojection errors.

Unlike BA algorithms, multi-view stereo algorithms are aimed at recovering scene information alone given fixed camera parameters.

DxO Optics Pro - Lens Distortion


In practice, we have found PMVS to be robust to errors in camera parameters as long as the image resolution matches the corresponding reprojection errors - that is, when features to be matched are roughly within two pixels of the corresponding 3D points.


> Algorithm

1) initializing feature correspondences
1-1) building image pyramids
1-2) for the given camera parameters, obtaining a conservative estimate of the expected reprojection error
1-3) sub-sampling

2) refining feature correspondences
2-1) determining a patch and the local image texture inside
2-2) optimizing the reference camera with the conjugate gradient method
2-3) removing outliers and updating the corresponding visibility information

3) updating the camera parameters with the SBA bundle adjustment software

4) repeat 1)-3) 4 times with the updated expected reprojection error and the fixed pyramid level







posted by maetel