The KITTI Vision Benchmark Suite

Extraction of depth from images is of great
importance for various computer vision
applications. Methods based on convolutional
neural networks are very accurate but have high
computation requirements, which can be achieved
with GPUs. However, GPUs are difficult to use on
devices with low power requirements like robots
and embedded systems. In this light, we propose a
stereo matching method appropriate for
applications in which limited computational and
energy resources are available. The algorithm is
based on a hierarchical representation of image
pairs which is used to restrict disparity search
range. We propose a cost function that takes into
account region contextual information and a cost
aggregation method that preserves disparity
borders. We tested the proposed method on the
Middlebury and KITTI benchmark data sets and on
the TrimBot2020 synthetic data. We achieved
accuracy and time efficiency results that show
that the method is suitable to be deployed on
embedded and robotics systems.

See paper.

@article{BRANDT2020,
title = "Efficient binocular stereo
correspondence matching with 1-D Max-Trees",
journal = "Pattern Recognition Letters",
year = "2020",
issn = "0167-8655",
doi =
"https://doi.org/10.1016/j.patrec.2020.02.019",
url =
"http://www.sciencedirect.com/science/article/pii
/S0167865520300581",
author = "Rafaël Brandt and Nicola Strisciuglio
and Nicolai Petkov and Michael H.F. Wilkinson",
keywords = "Stereo matching, Mathematical
morphology, Tree structures",
abstract = "Extraction of depth from images is of
great importance for various computer vision
applications. Methods based on convolutional
neural networks are very accurate but have high
computation requirements, which can be achieved
with GPUs. However, GPUs are difficult to use on
devices with low power requirements like robots
and embedded systems. In this light, we propose a
stereo matching method appropriate for
applications in which limited computational and
energy resources are available. The algorithm is
based on a hierarchical representation of image
pairs which is used to restrict disparity search
range. We propose a cost function that takes into
account region contextual information and a cost
aggregation method that preserves disparity
borders. We tested the proposed method on the
Middlebury and KITTI benchmark data sets and on
the TrimBot2020 synthetic data. We achieved
accuracy and time efficiency results that show
that the method is suitable to be deployed on
embedded and robotics systems."
}