Using printable speckle patterns for DIC: addressing potential and challenges S. J. C. Matthews 1a , D. A. Jesson 1 , M. J. Oldfield 1 and P. A. Smith 1 1 University of Surrey, Guildford, Surrey GU2 7XH a [email protected]Abstract Printing speckle patterns for digital image correlation (DIC) allows application of controlled patterns; this would limit variability between samples and user, whilst developing the field of multiscale speckles for global and detailed DIC assessment. The paper considers the positives for customisable fake tattoos as an application method, with a case study relating to human skin deformation and comparison against spray paint. Two current limitations (image compression and application upon curvature) are identified and potential solutions suggested. Introduction DIC has rapidly progressed, both in capability and number of users, since 1985 – the first occurrence of an 8-bit image being used to assess deformations using a speckle pattern 1,2 . Due to the technique’s non- contacting nature, which captures full-field deformation measurements, has uses for analysis of a large variety of materials, geometries, and scales 3–6 . The broad scope of the technique results in a spread of user competencies for a plethora of potential DIC applications; as such, in 2015, the non-profit organisation iDICs was formed for “training and educating users of DIC systems and the standardization of DIC practice for general applications“ 7,8 . One key aspect for DIC quality is the applied speckle pattern; non- optimal patterns risk decorrelation, aliasing, or the need for increasing subset size (lowering special resolution) to encapsulate enough speckles within the area 8–11 . Currently a common method for speckling components is with spray paint, but this is conducted “by eye” resulting in variation even between the same user’s specimens. A counter for the issue with variability is printing and adhesively applying desired patterns upon a sample; this has been successfully used for DIC analysis for flat, compact, specimens 12–14 , however an assessment into potential drawbacks of the application method is required for it to meet its potential. Case example: potential use of transfer paper for biomedical studies
Transcript
Using printable speckle patterns for DIC: addressing potential and challengesS. J. C. Matthews1a, D. A. Jesson1, M. J. Oldfield 1 and P. A. Smith1
Printing speckle patterns for digital image correlation (DIC) allows application of controlled patterns; this would limit variability between samples and user, whilst developing the field of multiscale speckles for global and detailed DIC assessment.
The paper considers the positives for customisable fake tattoos as an application method, with a case study relating to human skin deformation and comparison against spray paint. Two current limitations (image compression and application upon curvature) are identified and potential solutions suggested.
Introduction
DIC has rapidly progressed, both in capability and number of users, since 1985 – the first occurrence of an 8-bit image being used to assess deformations using a speckle pattern 1,2. Due to the technique’s non-contacting nature, which captures full-field deformation measurements, has uses for analysis of a large variety of materials, geometries, and scales3–6. The broad scope of the technique results in a spread of user competencies for a plethora of potential DIC applications; as such, in 2015, the non-profit organisation iDICs was formed for “training and educating users of DIC systems and the standardization of DIC practice for general applications“7,8.
One key aspect for DIC quality is the applied speckle pattern; non-optimal patterns risk decorrelation, aliasing, or the need for increasing subset size (lowering special resolution) to encapsulate enough speckles within the area8–11. Currently a common method for speckling components is with spray paint, but this is conducted “by eye” resulting in variation even between the same user’s specimens. A counter for the issue with variability is printing and adhesively applying desired patterns upon a sample; this has been successfully used for DIC analysis for flat, compact, specimens12–14, however an assessment into potential drawbacks of the application method is required for it to meet its potential.
Case example: potential use of transfer paper for biomedical studies
(a) (b) (c)
Figure 1: Example use of customisable transfer paper (for a human arm), with (a) and (b) showing a sample segment under tension and compression respectively. (c) provides the entire video via a QR code
The transfer paper used in previous work14 (manufactured by Sunnyscopa) is designed to produce customisable fake tattoos. Due to the primary design parameters, the material is FDA approved and therefore safe for use upon human skin; this results in the potential for biomedical analysis such as identifying high strain regions for “everyday” tasks. Figure 1 shows that the paper conforms to skin deformation and remains undamaged when uncontrollably stretched, folded, and placed in shear.
Speckle pattern comparison with spray paint
(a) (b)
Figure 2: Visual comparison of speckle patterns when applied using spray paint (left) and transfer paper (right) for a representative coupon
As previously stated spray paint is often the method for speckle pattern application which limits control of the process whilst potentially producing incorrect speckle diameters. Figure 2 qualitatively shows that for a representative coupon (flat surface area 165 x 155 mm), the spray paint pattern applied (a) has a lower contrast between speckle and base than for the printed example (b).
Figure 3: Histograms of pixel 8-bit greyscale values for the two speckle patterns in Figure 2. Vertical lines are coloured to represent the greyscale colour for 40 and 128 intensity
Due to the lack of contrast between spray and base, caused by poor spray paint application and insufficiently large speckle diameters, the 8-bit image for Figure 2(a) predominately ranges between greyscale values of 51 and 204 (as shown in Figure 3); the limited variation results in an increased probability of decorrelation because of limited clear speckles (which are identified as low greyscale value pixels). The transfer paper applied pattern opposes issues with paint, with the predominant greyscale value approximately 40, showing speckles are clearly identified in the image.
Compression of speckle patterns
Generation software is used to produce speckle patterns for transfer paper application, Correlated Solutions’ freeware SpeckleGen provides PDF files of the chosen pattern dependant on user defined variables for speckle diameter, density, and variation. When printing direct using portable document format (PDF) data is preserved15, so the the limiting factor for pattern quality relates to the number of dots per square inch (DPI) the printer is capable of; affordable home printers are capable of 1200 DPI1, resulting in 21 microns per dot, enough for the majority of DIC applications.
PDF
to J
PEG
con
vers
ion:
regi
on p
ixel
s
Speckle diameter (mm)
0.1 1
960
x 96
0
(21.
3 pi
xel p
er m
m)
1920
x 1
920
(42.
7 pi
xel p
er m
m)
3840
x 3
840
(85.
3 pi
xel p
er m
m)
Figure 4: Enlarged views of a single 0.1 mm and 1 mm diameter speckle, taken from a PDF pattern generated for a 45 x 45 mm region and compressed into JPEG format at three different compression ratios
Nevertheless, alternate styles and targets of speckle patterns require image formats for manipulation tools; an example being the addition of further details within a pattern, allowing for multiscale tracking using the same applied speckle16. For image manipulation compression of the PDF file is required. One method for PDF compression is the use of a standard set by the joint photographic experts group (JPEG), which: i) converts the file to YCbCr (Y: luminance; Cb: chroma blue; Cr: chroma red) format; ii) removes high luminosity information; iii) reduces image size; iv) applies a discrete cosine transform17. JPEG results in losses to image quality at locations of sharp contrast, which can be exaggerated with poor conversion parameters for a given feature size – an example of this is shown in Figure 4.
Figure 5: 8-bit greyscale intensity plot following the red arrow direction for all speckles shown in Figure 4, enlarges to a 400 x 400 pixel region
Visual assessment of Figure 4 clearly indicates that one conversion ratio is not suitable for all; respective extreme and partial pixilation visible for all 0.1 mm and two 1 mm diameter speckle, for chosen compressions. This is quantified in Figure 5 via a profile plot of the 8-bit greyscale intensity through a diagonal line (indicated by the red arrow across the speckle). Ideally the plot will drop from a value of 255 (white) to 0 (black) instantaneously, at approximately 110 pixels along the line; however, JPEG compression results in a region between 255 and 0 where greyscale value is constant – the length of the region provides an assessment of the extent of pixelization for the associated JPEG compression.
Application upon curvature
Figure 6: Speckle pattern applied using transfer paper upon a double-curved component, three regions are identified as problems relating to the application process: a) folds in the transfer paper during application; b) regions at which pattern coverage is lost; c)
overlapping regions resulting in changes to local speckle density
As indicated in Figure 1, the recommended transfer paper is as pliant as the skin it is bonded to, as well as capable of application and deformation for curved components; however, when transferring a 2D sheet of paper upon doubly curved objects there is the possibility for pattern discontinuities, such as the ones shown in Figure 6. Euler and Arcamedian spirals have shown that the use of geometrically spiralled nets can limit folds from original structural rigidity (Figure 6. ROI (a))- but this may have implications relating to gaps (Figure 6 ROI (b)) and overlaps (Figure 6, ROI (c)) within the pattern.
Conclusions
Speckle patterns using customisable transfer paper represent the possibility of a significant step in the quality control of DIC for a range of materials and applications – as indicated by the biomedical case study and comparison with spray paint.
Two draw-backs of the method have been assessed: JPEG compression, and application upon curvatures. Further assessment for JPEG compression will result in an understanding of memory vs speckle quality, thus providing recommendations for compression ratio at any given speckle diameter; whilst the production of mathematical nets, for component geometries, will limit folds, gaps, and overlaps, when applied upon the piece.
References
1. Chu, T. C., Ranson, W. F. & Sutton, M. A. Applications of digital-image-correlation techniques to experimental mechanics. Exp. Mech. 25, 232–244 (1985).
2. Sutton, M. A. Image-based Measurements in Solid Mechanics: A brief History, static and Dynamic Application Examples and Recent Developments. Showc. Lead. edge Exp. Tech. (2014). doi:10.1016/S1369-7021(10)70235-2
3. Sutton, M. A., Orteu, J.-J. & Schreier, H. Digital image correlation for shape and deformation measurements. (Springer, 2009). doi:10.1007/978-0-387-78747-3
4. Gencturk, B., Hossain, K., Kapadia, A., Labib, E. & Mo, Y. L. Use of digital image correlation technique in full-scale testing of prestressed concrete structures. Meas. J. Int. Meas. Confed. 47, 505–515 (2014).
5. Chen, Z. et al. Noninvasive, three-dimensional full-field body sensor for surface deformation monitoring of human body in vivo. J. Biomed. Opt. 22, 1 (2017).
6. Soltani, A. et al. Digital image correlation study of the deformation and functioning of the human heart during open-heart surgery. in Conference proceeding of the society for experimental mechanics series 19–27 (Springer, 2017). doi:https://doi.org/10.1007/978-3-319-63552-1_4
7. International Digital Image Society. iDICs homepage. (2015). Available at: http://idics.org/. (Accessed: 10th April 2019)
8. International Digital Image Society. A good practices guide for digital image correlation. (iDICs, 2018). doi:10.32720/idics/gpg.ed1
9. Correlated Solutions. Subset, step size and strain filter selection. Application notes (2016). Available at: https://www.correlatedsolutions.com/support/index.php?/Knowledgebase/Article/View/10/1/subset-step-size-and-strain-filter-selection. (Accessed: 12th December 2018)
11. Li, B. J., Wang, Q. B., Duan, D. P. & Chen, J. A. Using grey intensity adjustment strategy to enhance the measurement accuracy of digital image correlation considering the effect of intensity saturation. Opt. Lasers Eng. 104, 173–180 (2018).
12. Chen, Z., Quan, C., Zhu, F. & He, X. A method to transfer speckle patterns for digital image correlation. Meas. Sci. Technol. 26, (2015).
13. Ashrafi, M. & Tuttle, M. E. Measurement of Strain Gradients Using Digital Image Correlation by Applying Printed-Speckle Patterns. Exp. Tech. 40, 891–897 (2016).
14. Matthews, S. J. C. et al. A practical method for producing optimised and consistent speckle patterns for digital image correlation. Strain in review, (2019).
15. Adobe. What is PDF? (2018). Available at: https://acrobat.adobe.com/uk/en/acrobat/about-adobe-pdf.html. (Accessed: 29th June 2019)
16. Stoilov, G., Pashkouleva, D. & Kavardzhikov, V. Speckle patterns for multiscale monitoring of deformation. AIP Conf. Proc. 2075, (2019).
17. Pennebaker, W. B. & Mitchell, J. L. JPEG: still image compression standard (digital multimedia standards). (Springer, 1992).
