Stereo Image Datasets

Multiple-Viewpoint Video Sequences of Dynamic Scenes

Acquisition Process

Multi-view sequences were acquired with a rig of synchronized, progressive-scan Pulnix TMC9700 color cameras. Each camera was connected via RGBSync cables to a dedicated PC containing one multi-channel Matrox Meteor II framegrabber. Frames were grabbed uncompressed at a rate of 30 frames/second into the PC's internal memory.

The cameras allowed simultaneous observation of an approximately 30x30x30cm working volume, as shown below. Only seven of the rig's twelve cameras were used for acquiring the sequences.

All cameras were calibrated geometrically and radiometrically prior to image acquisition. Calibration was performed using the Vista implementation of Tsai's calibration algorithm. Reprojection accuracy within the working volume was approximately 0.5 pixels for the datasets below.

Scene illumination consisted of two 800W halogen point light sources whose 3D positions were recovered by adapting the method described by Bouguet.

Camera calibration and light source positions were unchanged for all sequences captured.

All sequences were acquired at the University of Rochester by Rodrigo Carceroni.

References

  • Rodrigo L. Carceroni and Kiriakos N. Kutulakos, "Multi-View Scene Capture by Surfel Sampling: From Video Streams to Non-Rigid 3D Motion, Shape and Reflectance," Int. J. Computer Vision, vol. 49, no. 2/3, pp. 175-214, 2002.
  • Rodrigo L. Carceroni and Kiriakos N. Kutulakos, "Multi-View Scene Capture by Surfel Sampling: From Video Streams to Non-Rigid 3D Motion, Shape and Reflectance," Proc. 8th Int. Conf. on Computer Vision, pp.60-67, 2001.

 

SEQUENCE #1: FLAG

Scene Data

[Note1: All X,Y,Z world coordinates are in mm]
[Note2: Bounding box information is approximate, i.e., some scene points may lie outside the box]

Approximate Scene Bounding Box Light Source #1 Light Source #2
Box center:
Box: Dimensions:
[60,-90,620]
[200,140,140]
[1173,-240.5,1870] [-1067,-248.5,1616]

Image Data

[Note1: MPEG links point to low-resolution previews of the full-resolution TIFF image sequences]
[Note2: Camera calibration parameters are in the text header of each Vista image. For an explanation of these parameters, see this page.]

Camera
#
First Frame Image
Sequences
First Frame in Vista Format with
Attached Calibration Parameters
1 TIFF

MPEG

Vista
2 TIFF

MPEG

Vista
3 TIFF

MPEG

Vista
4 TIFF

MPEG

Vista
5 TIFF

MPEG

Vista
6 TIFF

MPEG

Vista
7 TIFF

MPEG

Vista
 

SEQUENCE #2: NECK

Scene Data

[Note1: All X,Y,Z world coordinates are in mm]
[Note2: Bounding box information is approximate, i.e., some scene points may lie outside the box]

Approximate Scene Bounding Box Light Source #1 Light Source #2
Box center:
Box: Dimensions:
[60,-90,645]
[150,130,140]
[1173,-240.5,1870] [-1067,-248.5,1616]

Image Data

[Note1: MPEG links point to low-resolution previews of the full-resolution TIFF image sequences]
[Note2: Camera calibration parameters are in the text header of each Vista image. For an explanation of these parameters, see this page.]

Camera
#
First Frame Image
Sequences
First Frame in Vista Format with
Attached Calibration Parameters
1 TIFF

MPEG

Vista
2 TIFF

MPEG

Vista
3 TIFF

MPEG

Vista
4 TIFF

MPEG

Vista
5 TIFF

MPEG

Vista
6 TIFF

MPEG

Vista
7 TIFF

MPEG

Vista
 

SEQUENCE #3: VASE

Scene Data

[Note1: All X,Y,Z world coordinates are in mm]
[Note2: Bounding box information is approximate, i.e., some scene points may lie outside the box]

Approximate Scene Bounding Box Light Source #1 Light Source #2
Box center:
Box: Dimensions:
[43,-95,646]
[150,150,150]
[1173,-240.5,1870] [-1067,-248.5,1616]

Image Data

[Note1: MPEG links point to low-resolution previews of the full-resolution TIFF image sequences]
[Note2: Camera calibration parameters are in the text header of each Vista image. For an explanation of these parameters, see this page.]

Camera
#
First Frame Image
Sequences
First Frame in Vista Format with
Attached Calibration Parameters
1 TIFF

MPEG

Vista
2 TIFF

MPEG

Vista
3 TIFF

MPEG

Vista
4 TIFF

MPEG

Vista
5 TIFF

MPEG

Vista
6 TIFF

MPEG

Vista
7 TIFF

MPEG

Vista
 

SEQUENCE #4: SHIRT

Scene Data

[Note1: All X,Y,Z world coordinates are in mm]
[Note2: Bounding box information is approximate, i.e., some scene points may lie outside the box]

Approximate Scene Bounding Box
For Frame #19
Light Source #1 Light Source #2
Box center:
Box: Dimensions:
[45,-90,670]
[150,150,150]
[1173,-240.5,1870] [-1067,-248.5,1616]

Image Data

[Note1: MPEG links point to low-resolution previews of the full-resolution TIFF image sequences]
[Note2: Camera calibration parameters are in the text header of each Vista image. For an explanation of these parameters, see this page.]

Camera
#
First Frame Image
Sequences
First Frame in Vista Format with
Attached Calibration Parameters
1 TIFF

MPEG

Vista
2 TIFF

MPEG

Vista
3 TIFF

MPEG

Vista
4 TIFF

MPEG

Vista
5 TIFF

MPEG

Vista
6 TIFF

MPEG

Vista
7 TIFF

MPEG

Vista
 

 

 

Page last updated: Tuesday, September 06, 2005
For questions or comments send email to kyros@cs.toronto.edu