A Census-Based Stereo Vision Algorithm Using Modified Semi-Global Matching

This paper introduces a new segmentation-based approach
for disparity optimization in stereo vision. The main
contribution is a significant enhancement of the matching
quality at occlusions and textureless areas by segmenting
either the left color image or the calculated texture image.
The local cost calculation is done with a Census-based correlation
method and is compared with standard sum of absolute
differences. The confidence of a match is measured
and only non-confident or non-textured pixels are estimated
by calculating a disparity plane for the corresponding segment.
The quality of the local optimized matches is increased
by a modified Semi-Global Matching (SGM) step
with subpixel accuracy. In contrast to standard SGM, not
the whole image is used for disparity optimization but horizontal
stripes of the image. It is shown that this modification
significantly reduces the memory consumption by
nearly constant matching quality and thus enables embedded
realization. Using the Middlebury ranking as evaluation
criterion, it is shown that the proposed algorithm performs
well in comparison to the pure Census correlation.
It reaches a top ten rank if subpixel accuracy is supposed.
Furthermore, the matching quality of the algorithm, especially
of the texture-based plane fitting, is shown on two
real-world scenes where a significant enhancement could
be achieved.