Stereo vision enabling precise border localization

This page provides preliminary experimental results concerned with the ACCV 2007 paper "Stereo vision enabling precise border localization within a scanline optimization framework" (see [1]).
The algorithm relies on a novel adaptive matching cost function based on segmentation proposed in [2] and a scanline optimization framework for energy minimization.

Experimental results: stereo pairs with ground truth

Currently we have tested our algorithm [3] on the Middlebury dataset (availbale here). Detailed experimental results concerning algorithm [1], referred to as "SO+borders", and a complete comparison with other approaches is available at the Middlebury Stereo Evaluation web site.

Tsukuba Groundtruth Disparity map [1]

Venus Groundtruth Disparity map [1]

Teddy Groundtruth Disparity map [1]

Cones Groundtruth Disparity map [1]

References

[1] F. Tombari, S. Mattoccia, L. Di Stefano
“Segmentation-based adaptive support for accurate stereo correspondence "
IEEE Pacific-Rim Symposium on Image and Video Technology (PSIVT 2007)
December 17-19, 2007, Santiago, Chile
[Abstract] [Pdf] [Bibtex]

[2] S. Mattoccia, F. Tombari, L. Di Stefano,
“Stereo vision enabling precise border localization within a scanline optimization framework"
8th Asian Conference on Computer Vision (ACCV 2007)
November 18-22, 2007, Tokyo, Japan
[Abstract] [Pdf] [Bibtex]

[3] D. Scharstein and R. Szeliski.
"A taxonomy and evaluation of dense two-frame stereo correspondence algorithms"
IJCV 47(1/2/3):7-42, April-June 2002
Microsoft Research Technical Report MSR-TR-2001-81, November 2001