Digital Signal Processing Reference
In-Depth Information
Architectures for Stereo Vision
Christian Banz, Holger Blume, and Peter Pirsch
Abstract
Stereo vision is an elementary problem for many computer vision tasks.
It has been widely studied under the two aspects of increasing the quality of
the results and accelerating the computational processes. This chapter provides
theoretic background on stereo vision systems and discusses architectures and
implementations for real-time applications. In particular, the computationally most
intensive part, the stereo matching, is discussed on the example of one of the
leading algorithms, the semi-global matching (SGM). For this algorithm two
implementations are presented in detail on two of the most relevant platforms
for real-time image processing today: Field Programmable Gate Arrays (FPGAs)
and Graphics Processing Units (GPUs). Thus, the major differences in designing
parallelization techniques for extremely different image processing platforms are
being illustrated.
1
Introduction
The field of stereo vision is highly inspired by the capabilities of the human imaging
system. It encompasses all aspects of computer vision processing data from stereo
image pairs in one way or another. The goal is to estimate 3D information about
the observed scene, which can be used for a number of applications such as e.g.
distance measurement, 3D reconstruction, and arbitrary view interpolation. Crucial
for stereo vision is the task of stereo matching which identifies the projection points
of the same 3D real world point in both images of the stereo pair. The location
C. Banz (
)H.BlumeP.Pirsch
Institute of Microelectronic Systems, Leibniz University of Hannover Appelstr. 4,
30167 Hannover, Germany
