Some practical considerations in finite element-based digital image correlation

Handle URI:
http://hdl.handle.net/10754/564201
Title:
Some practical considerations in finite element-based digital image correlation
Authors:
Wang, Bo; Pan, Bing; Lubineau, Gilles ( 0000-0002-7370-6093 )
Abstract:
As an alternative to subset-based digital image correlation (DIC), finite element-based (FE-based) DIC method has gained increasing attention in the experimental mechanics community. However, the literature survey reveals that some important issues have not been well addressed in the published literatures. This work therefore aims to point out a few important considerations in the practical algorithm implementation of the FE-based DIC method, along with simple but effective solutions that can effectively tackle these issues. First, to better accommodate the possible intensity variations of the deformed images practically occurred in real experiments, a robust zero-mean normalized sum of squared difference criterion, instead of the commonly used sum of squared difference criterion, is introduced to quantify the similarity between reference and deformed elements in FE-based DIC. Second, to reduce the bias error induced by image noise and imperfect intensity interpolation, low-pass filtering of the speckle images with a 5×5 pixels Gaussian filter prior to correlation analysis, is presented. Third, to ensure the iterative calculation of FE-based DIC converges correctly and rapidly, an efficient subset-based DIC method, instead of simple integer-pixel displacement searching, is used to provide accurate initial guess of deformation for each calculation point. Also, the effects of various convergence criteria on the efficiency and accuracy of FE-based DIC are carefully examined, and a proper convergence criterion is recommended. The efficacy of these solutions is verified by numerical and real experiments. The results reveal that the improved FE-based DIC offers evident advantages over existing FE-based DIC method in terms of accuracy and efficiency. © 2015 Elsevier Ltd. All rights reserved.
KAUST Department:
Physical Sciences and Engineering (PSE) Division; Mechanical Engineering Program; Composite and Heterogeneous Material Analysis and Simulation Laboratory (COHMAS)
Publisher:
Elsevier BV
Journal:
Optics and Lasers in Engineering
Issue Date:
20-Apr-2015
DOI:
10.1016/j.optlaseng.2015.03.010
Type:
Article
ISSN:
01438166
Appears in Collections:
Articles; Physical Sciences and Engineering (PSE) Division; Mechanical Engineering Program

Full metadata record

DC FieldValue Language
dc.contributor.authorWang, Boen
dc.contributor.authorPan, Bingen
dc.contributor.authorLubineau, Gillesen
dc.date.accessioned2015-08-03T12:36:12Zen
dc.date.available2015-08-03T12:36:12Zen
dc.date.issued2015-04-20en
dc.identifier.issn01438166en
dc.identifier.doi10.1016/j.optlaseng.2015.03.010en
dc.identifier.urihttp://hdl.handle.net/10754/564201en
dc.description.abstractAs an alternative to subset-based digital image correlation (DIC), finite element-based (FE-based) DIC method has gained increasing attention in the experimental mechanics community. However, the literature survey reveals that some important issues have not been well addressed in the published literatures. This work therefore aims to point out a few important considerations in the practical algorithm implementation of the FE-based DIC method, along with simple but effective solutions that can effectively tackle these issues. First, to better accommodate the possible intensity variations of the deformed images practically occurred in real experiments, a robust zero-mean normalized sum of squared difference criterion, instead of the commonly used sum of squared difference criterion, is introduced to quantify the similarity between reference and deformed elements in FE-based DIC. Second, to reduce the bias error induced by image noise and imperfect intensity interpolation, low-pass filtering of the speckle images with a 5×5 pixels Gaussian filter prior to correlation analysis, is presented. Third, to ensure the iterative calculation of FE-based DIC converges correctly and rapidly, an efficient subset-based DIC method, instead of simple integer-pixel displacement searching, is used to provide accurate initial guess of deformation for each calculation point. Also, the effects of various convergence criteria on the efficiency and accuracy of FE-based DIC are carefully examined, and a proper convergence criterion is recommended. The efficacy of these solutions is verified by numerical and real experiments. The results reveal that the improved FE-based DIC offers evident advantages over existing FE-based DIC method in terms of accuracy and efficiency. © 2015 Elsevier Ltd. All rights reserved.en
dc.publisherElsevier BVen
dc.subjectDigital image correlationen
dc.subjectDisplacement measurementen
dc.subjectFinite elementen
dc.titleSome practical considerations in finite element-based digital image correlationen
dc.typeArticleen
dc.contributor.departmentPhysical Sciences and Engineering (PSE) Divisionen
dc.contributor.departmentMechanical Engineering Programen
dc.contributor.departmentComposite and Heterogeneous Material Analysis and Simulation Laboratory (COHMAS)en
dc.identifier.journalOptics and Lasers in Engineeringen
dc.contributor.institutionInstitute of Solid Mechanics, Beijing University of Aeronautics and AstronauticsBeijing, Chinaen
kaust.authorLubineau, Gillesen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.