Non-frontal Model Based Approach to Forensic Face Recognition

Research output: Contribution to conferenceAbstract

Abstract

In this paper, we propose a non-frontal model based approach which ensures that a face recognition system always gets to compare images having similar view (or pose). This requires a virtual suspect reference set that consists of non-frontal suspect images having pose similar to the surveillance view trace image. We apply the 3D model reconstruction followed by image synthesis approach to the frontal view mug shot images in the suspect reference set in order to create such a virtual suspect reference set. This strategy not only ensures a stable 3D face model reconstruction because of the relatively good quality mug shot suspect images but also provides a practical solution for forensic cases where the trace is often of very low quality. For most face recognition algorithms, the relative pose difference between the test and reference image is one of the major causes of severe degradation in recognition performance. Moreover, given appropriate training, comparing a pair of non-frontal images is no more difficult that comparing frontal view images.
LanguageEnglish
Number of pages1
StatePublished - 18 Jun 2012
EventICT - University of Twente, Enschede, Netherlands
Duration: 18 Jun 201218 Jun 2012

Conference

ConferenceICT
CountryNetherlands
CityEnschede
Period18/06/1218/06/12

Fingerprint

Face recognition
Degradation

Keywords

  • METIS-287887
  • IR-80774
  • forensic face recognition
  • EWI-21960
  • non-frontal model based approach
  • PIE $\cap$ MultiPIE problem
  • SCS-Safety
  • forensic problem

Cite this

@conference{10146205187a4a8092327a086b9cd01e,
title = "Non-frontal Model Based Approach to Forensic Face Recognition",
abstract = "In this paper, we propose a non-frontal model based approach which ensures that a face recognition system always gets to compare images having similar view (or pose). This requires a virtual suspect reference set that consists of non-frontal suspect images having pose similar to the surveillance view trace image. We apply the 3D model reconstruction followed by image synthesis approach to the frontal view mug shot images in the suspect reference set in order to create such a virtual suspect reference set. This strategy not only ensures a stable 3D face model reconstruction because of the relatively good quality mug shot suspect images but also provides a practical solution for forensic cases where the trace is often of very low quality. For most face recognition algorithms, the relative pose difference between the test and reference image is one of the major causes of severe degradation in recognition performance. Moreover, given appropriate training, comparing a pair of non-frontal images is no more difficult that comparing frontal view images.",
keywords = "METIS-287887, IR-80774, forensic face recognition, EWI-21960, non-frontal model based approach, PIE $\cap$ MultiPIE problem, SCS-Safety, forensic problem",
author = "A. Dutta and Veldhuis, {Raymond N.J.} and Spreeuwers, {Lieuwe Jan}",
note = "This poster received the first prize of {\^a}‚¬500 in the CTIT Poster Competition 2012. Out of the 35 submitted abstracts, 19 were selected to be presented at the poster competition. Updates available at: http://abhishekdutta.org/phd_ctit2012/. ; ICT : The Innovation Highway ; Conference date: 18-06-2012 Through 18-06-2012",
year = "2012",
month = "6",
day = "18",
language = "English",

}

Dutta, A, Veldhuis, RNJ & Spreeuwers, LJ 2012, 'Non-frontal Model Based Approach to Forensic Face Recognition' ICT, Enschede, Netherlands, 18/06/12 - 18/06/12, .

Non-frontal Model Based Approach to Forensic Face Recognition. / Dutta, A.; Veldhuis, Raymond N.J.; Spreeuwers, Lieuwe Jan.

2012. Abstract from ICT, Enschede, Netherlands.

Research output: Contribution to conferenceAbstract

TY - CONF

T1 - Non-frontal Model Based Approach to Forensic Face Recognition

AU - Dutta,A.

AU - Veldhuis,Raymond N.J.

AU - Spreeuwers,Lieuwe Jan

N1 - This poster received the first prize of €500 in the CTIT Poster Competition 2012. Out of the 35 submitted abstracts, 19 were selected to be presented at the poster competition. Updates available at: http://abhishekdutta.org/phd_ctit2012/.

PY - 2012/6/18

Y1 - 2012/6/18

N2 - In this paper, we propose a non-frontal model based approach which ensures that a face recognition system always gets to compare images having similar view (or pose). This requires a virtual suspect reference set that consists of non-frontal suspect images having pose similar to the surveillance view trace image. We apply the 3D model reconstruction followed by image synthesis approach to the frontal view mug shot images in the suspect reference set in order to create such a virtual suspect reference set. This strategy not only ensures a stable 3D face model reconstruction because of the relatively good quality mug shot suspect images but also provides a practical solution for forensic cases where the trace is often of very low quality. For most face recognition algorithms, the relative pose difference between the test and reference image is one of the major causes of severe degradation in recognition performance. Moreover, given appropriate training, comparing a pair of non-frontal images is no more difficult that comparing frontal view images.

AB - In this paper, we propose a non-frontal model based approach which ensures that a face recognition system always gets to compare images having similar view (or pose). This requires a virtual suspect reference set that consists of non-frontal suspect images having pose similar to the surveillance view trace image. We apply the 3D model reconstruction followed by image synthesis approach to the frontal view mug shot images in the suspect reference set in order to create such a virtual suspect reference set. This strategy not only ensures a stable 3D face model reconstruction because of the relatively good quality mug shot suspect images but also provides a practical solution for forensic cases where the trace is often of very low quality. For most face recognition algorithms, the relative pose difference between the test and reference image is one of the major causes of severe degradation in recognition performance. Moreover, given appropriate training, comparing a pair of non-frontal images is no more difficult that comparing frontal view images.

KW - METIS-287887

KW - IR-80774

KW - forensic face recognition

KW - EWI-21960

KW - non-frontal model based approach

KW - PIE $\cap$ MultiPIE problem

KW - SCS-Safety

KW - forensic problem

M3 - Abstract

ER -