Parallel Image Segmentation Using Reduction-Sweeps On Multicore Processors and GPUs
Programa de Engenharia de Sistemas de Computacao, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
IEEE 26th Conference on Graphics, Patterns and Images (SIBGRAPI), 2013
@InProceedings{FariasFariMarrClua:2013:PaImSe,
author={"Farias},
title={"Parallelimagesegmentationusingreduction-sweepsonmulticoreprocessorsandGPUs"},
booktitle={"Proceedings…"},
year={"2013"},
editor={"Boyer},
organization={"ConferenceonGraphics},
publisher={"IEEEComputerSociety’sConferencePublishingServices"},
address={"LosAlamitos"},
keywords={"Imagesegmentation},
conference-location={"Arequipa},
conference-year={"Aug.5-8},
language={"en"},
url={"http://urlib.net/sid.inpe.br/sibgrapi/2013/07.08.23.00"},
targetfile={"sibgrapi-camera-ready-no-bookmarks.pdf"},
urlaccessdate={"2013}
}
In this paper we introduce the Reduction Sweep algorithm, a novel graph-based image segmentation algorithm that is designed for easy parallelization. It is based on a clustering approach focusing on local image characteristics. Each pixel is compared with its neighbors in an implicitly independent manner, and those deemed sufficiently similar according to a color criterion are joined. We achieve fast execution times while still maintaining the visual quality of the results. The algorithm is presented in four different implementations: sequential CPU, parallel CPU, GPU, and hybrid CPU-GPU. We compare the execution times of the four versions with each other and with other closely related image segmentation algorithms.
July 19, 2013 by hgpu
