A Hardware Architecture for the Affine-Invariant Extension of SIFT

Joohyuk Yum, Chul Hee Lee, Jinwoo Park, Jin Sung Kim, Hyuk Jae Lee

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Affine-invariant extension of scale-invariant feature transform (ASIFT) algorithm requires a large amount of computation and memory access, and consequently, is hard to process in real time. In order to increase the operation speed of ASIFT algorithm, this paper proposes a new hardware architecture for the ASIFT algorithm. In order to reduce the memory access time, the affine transform is modified to allow external memory access in the raster-scan order with a little accuracy drop. In addition, image filtering with skewed kernel is proposed in order to reduce the memory space for image storage. Additional complexity reduction is attempted to reduce the number of simulated viewpoints. As a result, throughput of the affine transform module is increased to 325% and the proposed hardware processes a video graphics array-sized (640×480) video at 20 fps.

Original languageEnglish
Article number8010882
Pages (from-to)3251-3261
Number of pages11
JournalIEEE Transactions on Circuits and Systems for Video Technology
Volume28
Issue number11
DOIs
Publication statusPublished - 2018 Nov

Fingerprint

Affine transforms
Hardware
Data storage equipment
Throughput

All Science Journal Classification (ASJC) codes

  • Media Technology
  • Electrical and Electronic Engineering

Cite this

Yum, Joohyuk ; Lee, Chul Hee ; Park, Jinwoo ; Kim, Jin Sung ; Lee, Hyuk Jae. / A Hardware Architecture for the Affine-Invariant Extension of SIFT. In: IEEE Transactions on Circuits and Systems for Video Technology. 2018 ; Vol. 28, No. 11. pp. 3251-3261.
@article{8cb599e4e1834c768e79eae5121b24ee,
title = "A Hardware Architecture for the Affine-Invariant Extension of SIFT",
abstract = "Affine-invariant extension of scale-invariant feature transform (ASIFT) algorithm requires a large amount of computation and memory access, and consequently, is hard to process in real time. In order to increase the operation speed of ASIFT algorithm, this paper proposes a new hardware architecture for the ASIFT algorithm. In order to reduce the memory access time, the affine transform is modified to allow external memory access in the raster-scan order with a little accuracy drop. In addition, image filtering with skewed kernel is proposed in order to reduce the memory space for image storage. Additional complexity reduction is attempted to reduce the number of simulated viewpoints. As a result, throughput of the affine transform module is increased to 325{\%} and the proposed hardware processes a video graphics array-sized (640×480) video at 20 fps.",
author = "Joohyuk Yum and Lee, {Chul Hee} and Jinwoo Park and Kim, {Jin Sung} and Lee, {Hyuk Jae}",
year = "2018",
month = "11",
doi = "10.1109/TCSVT.2017.2740175",
language = "English",
volume = "28",
pages = "3251--3261",
journal = "IEEE Transactions on Circuits and Systems for Video Technology",
issn = "1051-8215",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "11",

}

A Hardware Architecture for the Affine-Invariant Extension of SIFT. / Yum, Joohyuk; Lee, Chul Hee; Park, Jinwoo; Kim, Jin Sung; Lee, Hyuk Jae.

In: IEEE Transactions on Circuits and Systems for Video Technology, Vol. 28, No. 11, 8010882, 11.2018, p. 3251-3261.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Hardware Architecture for the Affine-Invariant Extension of SIFT

AU - Yum, Joohyuk

AU - Lee, Chul Hee

AU - Park, Jinwoo

AU - Kim, Jin Sung

AU - Lee, Hyuk Jae

PY - 2018/11

Y1 - 2018/11

N2 - Affine-invariant extension of scale-invariant feature transform (ASIFT) algorithm requires a large amount of computation and memory access, and consequently, is hard to process in real time. In order to increase the operation speed of ASIFT algorithm, this paper proposes a new hardware architecture for the ASIFT algorithm. In order to reduce the memory access time, the affine transform is modified to allow external memory access in the raster-scan order with a little accuracy drop. In addition, image filtering with skewed kernel is proposed in order to reduce the memory space for image storage. Additional complexity reduction is attempted to reduce the number of simulated viewpoints. As a result, throughput of the affine transform module is increased to 325% and the proposed hardware processes a video graphics array-sized (640×480) video at 20 fps.

AB - Affine-invariant extension of scale-invariant feature transform (ASIFT) algorithm requires a large amount of computation and memory access, and consequently, is hard to process in real time. In order to increase the operation speed of ASIFT algorithm, this paper proposes a new hardware architecture for the ASIFT algorithm. In order to reduce the memory access time, the affine transform is modified to allow external memory access in the raster-scan order with a little accuracy drop. In addition, image filtering with skewed kernel is proposed in order to reduce the memory space for image storage. Additional complexity reduction is attempted to reduce the number of simulated viewpoints. As a result, throughput of the affine transform module is increased to 325% and the proposed hardware processes a video graphics array-sized (640×480) video at 20 fps.

UR - http://www.scopus.com/inward/record.url?scp=85028459814&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028459814&partnerID=8YFLogxK

U2 - 10.1109/TCSVT.2017.2740175

DO - 10.1109/TCSVT.2017.2740175

M3 - Article

AN - SCOPUS:85028459814

VL - 28

SP - 3251

EP - 3261

JO - IEEE Transactions on Circuits and Systems for Video Technology

JF - IEEE Transactions on Circuits and Systems for Video Technology

SN - 1051-8215

IS - 11

M1 - 8010882

ER -