Object recognition in compressed imagery

W. Brent Seales; Cheng J. Yuan; Wei Hu; Matthew D. Cutts

doi:10.1016/s0262-8856(97)00072-3

Object recognition in compressed imagery

W. Brent Seales, Cheng J. Yuan, Wei Hu, Matthew D. Cutts

Research output: Contribution to journal › Article › peer-review

19 Scopus citations

Abstract

Image-based applications can save time and space by operating on compressed data. The problem is that most mid- and high-level image operations, such as object recognition, are formulated as sequences of operations in the image domain. Such methods need direct access to pixel information as a starting point, but the pixel information in a compressed image stream is not immediately accessible. In this paper we show how to perform object recognition directly on compressed images (JPEG) and index frames from video streams (MPEG I-frames) without recovering explicit pixel information. The approach uses eigenvectors constructed from compressed image data. Our performance results show that a five-fold speedup can be gained by using compressed data.

Original language	English
Pages (from-to)	337-352
Number of pages	16
Journal	Image and Vision Computing
Volume	16
Issue number	5
DOIs	https://doi.org/10.1016/s0262-8856(97)00072-3
State	Published - Apr 27 1998

Bibliographical note

Funding Information:
The authors would like to thank the MIT Media Laboratory, the Olivetti Research Laboratory, and the Machine Vision group at the University of Essex for making their face recognition data and code publicly available. We thank Zhaojun Bai, Tom Hayden, Cid Srinivasan, Raphael Finkel and Mirek Truszczynski (University of Kentucky) for comments and valuable discussions. Finally, we gratefully acknowledge the support of the National Science Foundation for this project under grant number IRI-9308415.

Funding

The authors would like to thank the MIT Media Laboratory, the Olivetti Research Laboratory, and the Machine Vision group at the University of Essex for making their face recognition data and code publicly available. We thank Zhaojun Bai, Tom Hayden, Cid Srinivasan, Raphael Finkel and Mirek Truszczynski (University of Kentucky) for comments and valuable discussions. Finally, we gratefully acknowledge the support of the National Science Foundation for this project under grant number IRI-9308415.

Funders	Funder number
National Science Foundation (NSF)	IRI-9308415

Keywords

Compressed images
JPEG
MPEG
Object recognition

ASJC Scopus subject areas

Signal Processing
Computer Vision and Pattern Recognition

Access to Document

10.1016/s0262-8856(97)00072-3

Cite this

@article{6744f2eaf8c44e79861db1ad8de76538,

title = "Object recognition in compressed imagery",

abstract = "Image-based applications can save time and space by operating on compressed data. The problem is that most mid- and high-level image operations, such as object recognition, are formulated as sequences of operations in the image domain. Such methods need direct access to pixel information as a starting point, but the pixel information in a compressed image stream is not immediately accessible. In this paper we show how to perform object recognition directly on compressed images (JPEG) and index frames from video streams (MPEG I-frames) without recovering explicit pixel information. The approach uses eigenvectors constructed from compressed image data. Our performance results show that a five-fold speedup can be gained by using compressed data.",

keywords = "Compressed images, JPEG, MPEG, Object recognition",

author = "Seales, {W. Brent} and Yuan, {Cheng J.} and Wei Hu and Cutts, {Matthew D.}",

note = "Funding Information: The authors would like to thank the MIT Media Laboratory, the Olivetti Research Laboratory, and the Machine Vision group at the University of Essex for making their face recognition data and code publicly available. We thank Zhaojun Bai, Tom Hayden, Cid Srinivasan, Raphael Finkel and Mirek Truszczynski (University of Kentucky) for comments and valuable discussions. Finally, we gratefully acknowledge the support of the National Science Foundation for this project under grant number IRI-9308415.",

year = "1998",

month = apr,

day = "27",

doi = "10.1016/s0262-8856(97)00072-3",

language = "English",

volume = "16",

pages = "337--352",

number = "5",

}

TY - JOUR

T1 - Object recognition in compressed imagery

AU - Seales, W. Brent

AU - Yuan, Cheng J.

AU - Hu, Wei

AU - Cutts, Matthew D.

N1 - Funding Information: The authors would like to thank the MIT Media Laboratory, the Olivetti Research Laboratory, and the Machine Vision group at the University of Essex for making their face recognition data and code publicly available. We thank Zhaojun Bai, Tom Hayden, Cid Srinivasan, Raphael Finkel and Mirek Truszczynski (University of Kentucky) for comments and valuable discussions. Finally, we gratefully acknowledge the support of the National Science Foundation for this project under grant number IRI-9308415.

PY - 1998/4/27

Y1 - 1998/4/27

N2 - Image-based applications can save time and space by operating on compressed data. The problem is that most mid- and high-level image operations, such as object recognition, are formulated as sequences of operations in the image domain. Such methods need direct access to pixel information as a starting point, but the pixel information in a compressed image stream is not immediately accessible. In this paper we show how to perform object recognition directly on compressed images (JPEG) and index frames from video streams (MPEG I-frames) without recovering explicit pixel information. The approach uses eigenvectors constructed from compressed image data. Our performance results show that a five-fold speedup can be gained by using compressed data.

AB - Image-based applications can save time and space by operating on compressed data. The problem is that most mid- and high-level image operations, such as object recognition, are formulated as sequences of operations in the image domain. Such methods need direct access to pixel information as a starting point, but the pixel information in a compressed image stream is not immediately accessible. In this paper we show how to perform object recognition directly on compressed images (JPEG) and index frames from video streams (MPEG I-frames) without recovering explicit pixel information. The approach uses eigenvectors constructed from compressed image data. Our performance results show that a five-fold speedup can be gained by using compressed data.

KW - Compressed images

KW - JPEG

KW - MPEG

KW - Object recognition

UR - http://www.scopus.com/inward/record.url?scp=0032046533&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032046533&partnerID=8YFLogxK

U2 - 10.1016/s0262-8856(97)00072-3

DO - 10.1016/s0262-8856(97)00072-3

M3 - Article

AN - SCOPUS:0032046533

SN - 0262-8856

VL - 16

SP - 337

EP - 352

JO - Image and Vision Computing

JF - Image and Vision Computing

IS - 5

ER -

Object recognition in compressed imagery

Abstract

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this