3D object recognition using invariance

Andrew Zisserman, David Forsyth, Joseph Mundy, Charlie Rothwell, Jane Liu, Nic Pillow

Research output: Contribution to journalArticle

Abstract

The systems and concepts described in this paper document the evolution of the geometric invariance approach to object recognition over the last five years. Invariance overcomes one of the fundamental difficulties in recognising objects from images: that the appearance of an object depends on viewpoint. This problem is entirely avoided if the geometric description is unaffected by the imaging transformation. Such invariant descriptions can be measured from images without any prior knowledge of the position, orientation and calibration of the camera. These invariant measurements can be used to index a library of object models for recognition and provide a principled basis for the other stages of the recognition process such as feature grouping and hypothesis verification. Object models can be acquired directly from images, allowing efficient construction of model libraries without manual intervention. A significant part of the paper is a summary of recent results on the construction of invariants for 3D objects from a single perspective view. A proposed recognition architecture is described which enables the integration of multiple general object classes and provides a means for enforcing global scene consistency. Various criticisms of the invariant approach are articulated and addressed.

Original languageEnglish (US)
Pages (from-to)239-288
Number of pages50
JournalArtificial Intelligence
Volume78
Issue number1-2
DOIs
StatePublished - Oct 1995
Externally publishedYes

Fingerprint

Object recognition
Invariance
Cameras
Calibration
Imaging techniques
Object Recognition
grouping
criticism
knowledge

ASJC Scopus subject areas

  • Language and Linguistics
  • Linguistics and Language
  • Artificial Intelligence

Cite this

Zisserman, A., Forsyth, D., Mundy, J., Rothwell, C., Liu, J., & Pillow, N. (1995). 3D object recognition using invariance. Artificial Intelligence, 78(1-2), 239-288. https://doi.org/10.1016/0004-3702(95)00023-2

3D object recognition using invariance. / Zisserman, Andrew; Forsyth, David; Mundy, Joseph; Rothwell, Charlie; Liu, Jane; Pillow, Nic.

In: Artificial Intelligence, Vol. 78, No. 1-2, 10.1995, p. 239-288.

Research output: Contribution to journalArticle

Zisserman, A, Forsyth, D, Mundy, J, Rothwell, C, Liu, J & Pillow, N 1995, '3D object recognition using invariance', Artificial Intelligence, vol. 78, no. 1-2, pp. 239-288. https://doi.org/10.1016/0004-3702(95)00023-2
Zisserman A, Forsyth D, Mundy J, Rothwell C, Liu J, Pillow N. 3D object recognition using invariance. Artificial Intelligence. 1995 Oct;78(1-2):239-288. https://doi.org/10.1016/0004-3702(95)00023-2
Zisserman, Andrew ; Forsyth, David ; Mundy, Joseph ; Rothwell, Charlie ; Liu, Jane ; Pillow, Nic. / 3D object recognition using invariance. In: Artificial Intelligence. 1995 ; Vol. 78, No. 1-2. pp. 239-288.
@article{59a342d0c0f143fd81ac5278e86d2f5c,
title = "3D object recognition using invariance",
abstract = "The systems and concepts described in this paper document the evolution of the geometric invariance approach to object recognition over the last five years. Invariance overcomes one of the fundamental difficulties in recognising objects from images: that the appearance of an object depends on viewpoint. This problem is entirely avoided if the geometric description is unaffected by the imaging transformation. Such invariant descriptions can be measured from images without any prior knowledge of the position, orientation and calibration of the camera. These invariant measurements can be used to index a library of object models for recognition and provide a principled basis for the other stages of the recognition process such as feature grouping and hypothesis verification. Object models can be acquired directly from images, allowing efficient construction of model libraries without manual intervention. A significant part of the paper is a summary of recent results on the construction of invariants for 3D objects from a single perspective view. A proposed recognition architecture is described which enables the integration of multiple general object classes and provides a means for enforcing global scene consistency. Various criticisms of the invariant approach are articulated and addressed.",
author = "Andrew Zisserman and David Forsyth and Joseph Mundy and Charlie Rothwell and Jane Liu and Nic Pillow",
year = "1995",
month = "10",
doi = "10.1016/0004-3702(95)00023-2",
language = "English (US)",
volume = "78",
pages = "239--288",
journal = "Artificial Intelligence",
issn = "0004-3702",
publisher = "Elsevier",
number = "1-2",

}

TY - JOUR

T1 - 3D object recognition using invariance

AU - Zisserman, Andrew

AU - Forsyth, David

AU - Mundy, Joseph

AU - Rothwell, Charlie

AU - Liu, Jane

AU - Pillow, Nic

PY - 1995/10

Y1 - 1995/10

N2 - The systems and concepts described in this paper document the evolution of the geometric invariance approach to object recognition over the last five years. Invariance overcomes one of the fundamental difficulties in recognising objects from images: that the appearance of an object depends on viewpoint. This problem is entirely avoided if the geometric description is unaffected by the imaging transformation. Such invariant descriptions can be measured from images without any prior knowledge of the position, orientation and calibration of the camera. These invariant measurements can be used to index a library of object models for recognition and provide a principled basis for the other stages of the recognition process such as feature grouping and hypothesis verification. Object models can be acquired directly from images, allowing efficient construction of model libraries without manual intervention. A significant part of the paper is a summary of recent results on the construction of invariants for 3D objects from a single perspective view. A proposed recognition architecture is described which enables the integration of multiple general object classes and provides a means for enforcing global scene consistency. Various criticisms of the invariant approach are articulated and addressed.

AB - The systems and concepts described in this paper document the evolution of the geometric invariance approach to object recognition over the last five years. Invariance overcomes one of the fundamental difficulties in recognising objects from images: that the appearance of an object depends on viewpoint. This problem is entirely avoided if the geometric description is unaffected by the imaging transformation. Such invariant descriptions can be measured from images without any prior knowledge of the position, orientation and calibration of the camera. These invariant measurements can be used to index a library of object models for recognition and provide a principled basis for the other stages of the recognition process such as feature grouping and hypothesis verification. Object models can be acquired directly from images, allowing efficient construction of model libraries without manual intervention. A significant part of the paper is a summary of recent results on the construction of invariants for 3D objects from a single perspective view. A proposed recognition architecture is described which enables the integration of multiple general object classes and provides a means for enforcing global scene consistency. Various criticisms of the invariant approach are articulated and addressed.

UR - http://www.scopus.com/inward/record.url?scp=0029388249&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029388249&partnerID=8YFLogxK

U2 - 10.1016/0004-3702(95)00023-2

DO - 10.1016/0004-3702(95)00023-2

M3 - Article

AN - SCOPUS:0029388249

VL - 78

SP - 239

EP - 288

JO - Artificial Intelligence

JF - Artificial Intelligence

SN - 0004-3702

IS - 1-2

ER -