On Biases in Displacement Estimation for Image Registration, with a Focus on Photomechanics

Sur, Frédéric; Blaysat, Benoît; Grédiac, Michel

doi:10.1007/s10851-021-01032-4

Cited by 18 publications

(17 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This effect can however be considered as negligible when the actual displacement and strain fields only gently evolve over the surface of the specimen, [37] which is the case here despite the presence of a hole. The pattern‐induced bias recently described in Fayad et al [38] and Lehoucq et al [39] is negligible for periodic patterns such as checkerboards [40] . It is also worth remembering that other sources of error such as distortion, non‐parallelism between sensor and specimen or parasitic out‐of‐plane movement may also affect the final results, but they are not discussed in this study.…”

Section: Resultsmentioning

confidence: 77%

Using laser marking to engrave optimal patterns for in‐plane displacement and strain measurement

Bouyra

Blaysat

Chanal

et al. 2021

Strain

Self Cite

View full text Add to dashboard Cite

The checkerboard constitutes the best pattern for in‐plane displacement and strain measurement because it maximizes image gradient. The use of laser marking to deposit such a pattern on flat surfaces is investigated in this paper. Optimal settings for the parameters influencing the quality of the pattern are given. This pattern being periodic, the images are processed with the localized spectrum analysis, which is one of the spectral techniques used to process such images. It is shown that the metrological performance in terms of measurement resolution is equivalent to the one obtained with transferred checkerboards. Compared to the classic transfer technique generally used to deposit such patterns, the benefit of using laser marking is its ease of use, a much higher print speed and the fact that a thin coat of white paint instead of a thicker layer of white adhesive is employed to ensure a good contrast in the images. Various examples of strain measurements are given, in particular the measurement of the strain field around a knot embedded in a wood specimen subjected to a tensile test.

show abstract

Section: Resultsmentioning

confidence: 77%

Using laser marking to engrave optimal patterns for in‐plane displacement and strain measurement

Bouyra

Blaysat

Chanal

et al. 2021

Strain

Self Cite

View full text Add to dashboard Cite

show abstract

“…The value of d must be as small as possible to reflect a small spatial resolution, thus the ability of the measuring technique of interest to distinguish close features in displacement and strain maps, and return a value of the displacements and strains in these regions with a small bias. In certain cases, the displacement resolution can be predicted theoretically from the transfer function of the filter associated to the technique (Savitzky-Golay filter for DIC [5,6], Gaussian filter for the Localized Spectrum Analysis [48]).…”

Section: Spatial Resolutionmentioning

confidence: 99%

“…It manifests itself by the presence of random fluctuations in the displacement and strain maps. These random fluctuations shall not be confused with sensor noise propagation since it is due to different causes, mainly the image gradient distribution and the difference between the true displacement field and its local approximation by subset shape functions, see [6] where a model for this phenomenon is proposed. These spatial fluctuations are randomly distributed because speckle patterns are randomly distributed.…”

Section: Pattern-induced Biasmentioning

confidence: 99%

“…Another drawback is the fact that DIC acts as a low-pass filter, which causes the retrieved displacement and strain fields to be blurred. Indeed, it is shown in [5,6] that the displacement field rendered by a DIC system is not the actual one, but regardless of noise, the actual one convolved by a Savitzky-Golay filter. This makes DIC to be unable to correctly render displacement or strain fields featuring high spatial frequencies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

When Deep Learning Meets Digital Image Correlation

Boukhtache

Abdelouahab

Berry

et al. 2021

Optics and Lasers in Engineering

View full text Add to dashboard Cite

“…Next, deconvolution was performed according to the approach proposed by Grediac et al [40]. This deconvolution algorithm relies on the fact that DIC displacements correspond, when omitting the effect of the pattern [60], to the physical displacement convoluted with a Savitzky-Golary kernel [61,62], which can be directly related to the subset size, shape and order. The deconvolution consists of correcting the higher order spatial derivative terms, which are computed by differentiation.…”

Section: High-resolution Dic Studymentioning

confidence: 99%

A Nanomechanical Testing Framework Yielding Front&Rear-Sided, High-Resolution, Microstructure-Correlated SEM-DIC Strain Fields

et al. 2022

View full text Add to dashboard Cite

Background The continuous development of new multiphase alloys with improved mechanical properties requires quantitative microstructure-resolved observation of the nanoscale deformation mechanisms at, e.g., multiphase interfaces. This calls for a combinatory approach beyond advanced testing methods such as microscale strain mapping on bulk material and micrometer sized deformation tests of single grains. Objective We propose a nanomechanical testing framework that has been carefully designed to integrate several state-of-the-art testing and characterization methods. Methods (i) Well-defined nano-tensile testing of carefully selected and isolated multiphase specimens, (ii) front&rear-sided SEM-EBSD microstructural characterization combined with front&rear-sided in-situ SEM-DIC testing at very high resolution enabled by a recently developed InSn nano-DIC speckle pattern, (iii) optimized DIC strain mapping aided by application of SEM scanning artefact correction and DIC deconvolution for improved spatial resolution, (iv) a novel microstructure-to-strain alignment framework to deliver front&rear-sided, nanoscale, microstructure-resolved strain fields, and (v) direct comparison of microstructure, strain and SEM-BSE damage maps in the deformed configuration. Results Demonstration on a micrometer-sized dual-phase steel specimen, containing an incompatible ferrite-martensite interface, shows how the nanoscale deformation mechanisms can be unraveled. Discrete lath-boundary-aligned martensite strain localizations transit over the interface into diffuse ferrite plasticity, revealed by the nanoscale front&rear-sided microstructure-to-strain alignment and optimization of DIC correlations. Conclusions The proposed testing and alignment framework yields front&rear-sided aligned microstructure and strain fields providing 3D interpretation of the deformations and opening new opportunities for unprecedented validation of advanced multiphase simulations.

show abstract

On Biases in Displacement Estimation for Image Registration, with a Focus on Photomechanics

Cited by 18 publications

References 45 publications

Using laser marking to engrave optimal patterns for in‐plane displacement and strain measurement

Using laser marking to engrave optimal patterns for in‐plane displacement and strain measurement

When Deep Learning Meets Digital Image Correlation

A Nanomechanical Testing Framework Yielding Front&Rear-Sided, High-Resolution, Microstructure-Correlated SEM-DIC Strain Fields

Contact Info

Product

Resources

About