Parallel Implementations of the CSBP Stereo Vision Algorithm

Author

Veronese, Lucas ; Junior, Lauro J Lyrio ; de Oliveira Neto, Jorcy ; Forechi, Avelino ; Badue, Claudine ; De Souza, Alberto F.

Author_Institution

Dept. de Inf., Univ. Fed. do Espfrito Santo, Vitoria, Brazil

fYear

2011

fDate

26-29 Oct. 2011

Firstpage

12

Lastpage

12

Abstract

Summary form only given. We developed two parallel versions of the Constant Space Belief Propagation algorithm (CSBP - one of the best stereo algorithms currently known): one in OpenMP and one in C+CUDA. For images with 640x480 pixels, the sequential version has a performance of 1.16 frames per second (FPS), the OpenMP parallel version has a performance of 3.7 FPS, while the C+CUDA version has a performance of 17.3 FPS in high-performance desktop machines. These results are important because they enable the implementation of autonomous vehicles with sensors like camera, which is one of the objectives of a PRONEX project being currently developed in LCAD-DI/UFES. One of the goals of this project is to implement an autonomous vehicle from a commercial automobile.

Keywords

C language; parallel architectures; stereo image processing; traffic engineering computing; C+CUDA; CSBP stereo vision algorithm; LCAD-DI-UFES; OpenMP parallel version; PRONEX project; autonomous vehicles; commercial automobile; constant space belief propagation algorithm; high-performance desktop machines; parallel implementations; sensors; Automobiles; Belief propagation; Cameras; Graphics processing unit; Mobile robots; Sensors; Stereo vision; Artificial Stereo Vison; Compute Unified Device Architecture; Constant Space Belief Propagation; OpenMP;

fLanguage

English

Publisher

ieee

Conference_Titel

Sistemas Computacionais (WSCAD-SSC), 2011 Simpasio em

Conference_Location

Vitoria

Print_ISBN

978-1-4673-0303-3

Type

conf

DOI

10.1109/WSCAD-SSC.2011.14

Filename

6113024