Abstract
We present a robust and accurate 3D registration method for a dense sequence of depth images taken from unknown viewpoints. Our method simultaneously estimates multiple extrinsic parameters of the depth images to obtain a registered full 3D model of the scanned scene. By arranging the depth measurements in a matrix form, we formulate the problem as a simultaneous estimation of multiple extrinsics and a low-rank matrix, which corresponds to the aligned depth images as well as a sparse error matrix. Unlike previous approaches that use sequential or heuristic global registration approaches, our solution method uses an advanced convex optimization technique for obtaining a robust solution via rank minimization. To achieve accurate computation, we develop a depth projection method that has minimum sensitivity to sampling by reading projected depth values in the input depth images. We demonstrate the effectiveness of the proposed method through extensive experiments and compare it with previous standard techniques.
| Original language | English |
|---|---|
| Title of host publication | Proceedings - 2nd Joint 3DIM/3DPVT Conference |
| Subtitle of host publication | 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012 |
| Pages | 33-40 |
| Number of pages | 8 |
| DOIs | |
| Publication status | Published - Dec 1 2012 |
| Event | 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012 - Zurich, Switzerland Duration: Oct 13 2012 → Oct 15 2012 |
Publication series
| Name | Proceedings - 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012 |
|---|
Other
| Other | 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012 |
|---|---|
| Country | Switzerland |
| City | Zurich |
| Period | 10/13/12 → 10/15/12 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Computer Vision and Pattern Recognition
- Modelling and Simulation
Cite this
Robust simultaneous 3D registration via rank minimization. / Thomas, Diego Gabriel Francis; Matsushita, Yasuyuki; Sugimoto, Akihiro.
Proceedings - 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012. 2012. p. 33-40 6374974 (Proceedings - 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Robust simultaneous 3D registration via rank minimization
AU - Thomas, Diego Gabriel Francis
AU - Matsushita, Yasuyuki
AU - Sugimoto, Akihiro
PY - 2012/12/1
Y1 - 2012/12/1
N2 - We present a robust and accurate 3D registration method for a dense sequence of depth images taken from unknown viewpoints. Our method simultaneously estimates multiple extrinsic parameters of the depth images to obtain a registered full 3D model of the scanned scene. By arranging the depth measurements in a matrix form, we formulate the problem as a simultaneous estimation of multiple extrinsics and a low-rank matrix, which corresponds to the aligned depth images as well as a sparse error matrix. Unlike previous approaches that use sequential or heuristic global registration approaches, our solution method uses an advanced convex optimization technique for obtaining a robust solution via rank minimization. To achieve accurate computation, we develop a depth projection method that has minimum sensitivity to sampling by reading projected depth values in the input depth images. We demonstrate the effectiveness of the proposed method through extensive experiments and compare it with previous standard techniques.
AB - We present a robust and accurate 3D registration method for a dense sequence of depth images taken from unknown viewpoints. Our method simultaneously estimates multiple extrinsic parameters of the depth images to obtain a registered full 3D model of the scanned scene. By arranging the depth measurements in a matrix form, we formulate the problem as a simultaneous estimation of multiple extrinsics and a low-rank matrix, which corresponds to the aligned depth images as well as a sparse error matrix. Unlike previous approaches that use sequential or heuristic global registration approaches, our solution method uses an advanced convex optimization technique for obtaining a robust solution via rank minimization. To achieve accurate computation, we develop a depth projection method that has minimum sensitivity to sampling by reading projected depth values in the input depth images. We demonstrate the effectiveness of the proposed method through extensive experiments and compare it with previous standard techniques.
UR - http://www.scopus.com/inward/record.url?scp=84872050273&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84872050273&partnerID=8YFLogxK
U2 - 10.1109/3DIMPVT.2012.15
DO - 10.1109/3DIMPVT.2012.15
M3 - Conference contribution
SN - 9780769548739
T3 - Proceedings - 2nd Joint 3DIM/3DPVT Conference: 3D Imaging, Modeling, Processing, Visualization and Transmission, 3DIMPVT 2012
SP - 33
EP - 40
BT - Proceedings - 2nd Joint 3DIM/3DPVT Conference
ER -