Digital Image Correlation (DIC) is a powerful tool used to evaluate displacements and deformations in a non-intrusive manner. By comparing two images, one of the undeformed reference state of a specimen and another of the deformed target state, the relative displacement between those two states is determined. DIC is well known and often used for post-processing analysis of in-plane displacements and deformation of specimen. Increasing the analysis speed to enable real-time DIC analysis will be beneficial and extend the field of use of this technique. Here we tested several combinations of the most common DIC methods in combination with different parallelization approaches in MATLAB and evaluated their performance to determine whether real-time analysis is possible with these methods. To reflect improvements in computing technology different hardware settings were also analysed. We found that implementation problems can reduce the efficiency of a theoretically superior algorithm such that it becomes practically slower than a sub-optimal algorithm. The Newton-Raphson algorithm in combination with a modified Particle Swarm algorithm in parallel image computation was found to be most effective. This is contrary to theory, suggesting that the inverse-compositional Gauss-Newton algorithm is superior. As expected, the Brute Force Search algorithm is the least effective method. We also found that the correct choice of parallelization tasks is crucial to achieve improvements in computing speed. A poorly chosen parallelisation approach with high parallel overhead leads to inferior performance. Finally, irrespective of the computing mode the correct choice of combinations of integer-pixel and sub-pixel search algorithms is decisive for an efficient analysis. Using currently available hardware real-time analysis at high framerates remains an aspiration.
Deep Dive into Significance of parallel computing on the performance of Digital Image Correlation algorithms in MATLAB.
Digital Image Correlation (DIC) is a powerful tool used to evaluate displacements and deformations in a non-intrusive manner. By comparing two images, one of the undeformed reference state of a specimen and another of the deformed target state, the relative displacement between those two states is determined. DIC is well known and often used for post-processing analysis of in-plane displacements and deformation of specimen. Increasing the analysis speed to enable real-time DIC analysis will be beneficial and extend the field of use of this technique. Here we tested several combinations of the most common DIC methods in combination with different parallelization approaches in MATLAB and evaluated their performance to determine whether real-time analysis is possible with these methods. To reflect improvements in computing technology different hardware settings were also analysed. We found that implementation problems can reduce the efficiency of a theoretically superior algorithm such th
1
Significance of parallel computing on the performance of Digital Image Correla-
tion algorithms in MATLAB
Andreas Thoma1,* and Sridhar Ravi2
1 Royal Melbourne Institute of Technology (RMIT University), Melbourne, Australia and FH Aachen, Aachen, Germany
2 Royal Melbourne Institute of Technology (RMIT University), Melbourne, Australia
Abstract
Digital Image Correlation (DIC) is a powerful tool used to
evaluate displacements and deformations in a non-intru-
sive manner. By comparing two images, one of the unde-
formed reference state of a specimen and another of the
deformed target state, the relative displacement be-
tween those two states is determined. DIC is well known
and often used for post-processing analysis of in-plane
displacements and deformation of specimen. Increasing
the analysis speed to enable real-time DIC analysis will be
beneficial and extend the field of use of this technique.
Here we tested several combinations of the most com-
mon DIC methods in combination with different parallel-
ization approaches in MATLAB and evaluated their per-
formance to determine whether real-time analysis is pos-
sible with these methods. To reflect improvements in
computing technology different hardware settings were
also analysed. We found that implementation problems
can reduce the efficiency of a theoretically superior algo-
rithm such that it becomes practically slower than a sub-
optimal algorithm. The Newton-Raphson algorithm in
combination with a modified Particle Swarm algorithm in
parallel image computation was found to be most effec-
tive. This is contrary to theory, suggesting that the in-
verse-compositional Gauss-Newton algorithm is supe-
rior. As expected, the Brute Force Search algorithm is the
least effective method. We also found that the correct
choice of parallelization tasks is crucial to achieve im-
provements in computing speed. A poorly chosen paral-
lelisation approach with high parallel overhead leads to
inferior performance. Finally, irrespective of the compu-
ting mode the correct choice of combinations of integer-
pixel and sub-pixel search algorithms is decisive for an ef-
ficient analysis. Using currently available hardware real-
time analysis at high framerates remains an aspiration.
Keywords: Digital Image Correlation, Real-Time Pro-
cessing, Newton-Raphson Method, Particle Swarm Optimi-
sation, Inverse-Compositional Gauss-Newton Method,
Parallel Computation
1
Introduction
In many different engineering applications the measure-
ment of displacements and deformations play an im-
portant role [1]. Due to limitations of invasive methods for
strain estimations, there exists a strong motivation for the
development of contactless measurement techniques. Op-
tical based strain measurement systems have a special im-
portance in the field of measurement systems as they offer
a potential, non-invasive, high throughput strain measure-
ment system. Therefore, over the last decade several dif-
ferent measurement systems for optical displacement
measurement have been developed [2, 3]. Digital Image
Correlation (DIC) gained a lot of popularity not only be-
cause of its simplicity and accuracy [4] but also because of
its robustness and wide field of applications [5]. After this
method was first published by Peters and Ranson [6] in the
early 1980s, the DIC method has been continuously im-
proved by different researchers, while the basic principle
remained the same. Firstly, accurately time resolved im-
ages of a specimen undergoing deformation are acquired.
The subsequent analysis is conducted between image pairs
separated over time where one of the images is considered
as the reference state while the other represents the de-
formed state of the specimen. Finally, the image pairs are
compared by dividing the reference image into sub-images
and searching for the same sub-image in the image of the
deformed state. By estimating the inter-image positions of
the sub-image pairs, a local displacement is determined.
2
The method of dividing the image into sub-images, search-
ing between images and processing routines have under-
gone many changes and improvements over the years to
develop accurate techniques for use in a variety of engi-
neering fields [7]. However, one of the main disadvantages
of DIC is its high computational burden and the relatively
slow image processing time, which imposes significant lim-
itations on the frame rate of image acquisition, and post-
processing of images. Additionally, DIC is usually restricted
to 2D analyses because of its high processing times [5].
Here we tested the most common computational strate-
gies used in DIC and evaluated their performance under se-
rial and parallel computational implementation in
MATLAB. MATLAB is a computing environment commonly
used by engineers, researchers and economists [8].
MATLAB com
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
