Roger Groves

This presentation describes optical and ultrasonic measuring principles for robust and reliable flaw detection, localization and characterization in non-destructive testing for composite manufacturing. The relevant manufacturing tasks as well as the measurement of strain, characterization of vibration and measurement of associated parameters such as shape and position are also presented. The lecture covers the application of optical metrology, fibre optics sensors, spectral imaging and ultrasound measuring techniques for non-destructive testing of aerospace materials and structures and civil engineering objects.

ABSTRACT The development of smart materials for embedding in aerospace composites provides enhanced functionality for future aircraft structures. Critical flight conditions like icing of the leading edges can affect the aircraft functionality and controllability. Hence, anti-icing and de-icing capabilities are used. In case of leading edges made of fibre metal laminates heater elements can be embedded between composite layers. However this local heating causes strains and stresses in the structure due to the different thermal expansion coefficients of the different laminated materials. In order to characterize the structural behaviour during thermal loading full-field strain and shape measurement can be used. In this research, a shearography instrument with three spatially-distributed shearing cameras is used to measure surface displacement gradients which give a quantitative estimation of the in- and out-of-plane surface strain components. For the experimental part, two GLARE (Glass Laminate Aluminum Reinforced Epoxy) specimens with six different embedded copper heater elements were manufactured: two copper mesh shapes (straight and S-shape), three connection techniques (soldered, spot welded and overlapped) and one straight heater element with delaminations. The surface strain behaviour of the specimens due to thermal loading was measured and analysed. The comparison of the connection techniques of heater element parts showed that the overlapped connection has the smallest effect on the surface strain distribution. Furthermore, the possibility of defect detection and defect depth characterisation close to the heater elements was also investigated.

Among the various types of wall decorations used in the Western world during the 16th to 18th centuries, gilt leather belonged to the most fashionable and costly ones. Despite its appearance, it is not real gold that creates the golden shine, but typically a silver leaf which is coated with an orange-brown lacquer to obtain the characteristic golden lustre. Due to its fragile character, only a fraction of the gilt leathers have survived in situ or in museum collections. Even though some of these gilt leathers are hanging in prominent locations, it is both nationally and internationally an overlooked part of our cultural heritage. Conservation work on gilt leather is a real challenge due to the complex nature of the material, historical craft-led conservation practices and the lack of systematic analytical research of degradation processes. This white paper presents an overview of the current state of the art in gilt leather research and conservation, and identifies future research directions. It is the result of a cross-disciplinary collaborative research project on gilt leather started in 2015. As part of this an international group of experts in gilt leather met at a Symposium in Maastricht, The Netherlands (March 2016). One of the outcomes of this meeting, covering the topics of art history, conservation, material dynamics and diagnostics, was a research agenda for gilt leather 2017-2025.

-- Note: would you rather watch online? http://youtu.be/x5-FF7ayfKQ Phase unwrapping is one of the key steps of interferogram analysis, and its accuracy relies primarily on the correct identification of phase discontinuities. This can be especially challenging for inherently noisy phase fields, such as those produced through shearography and other speckle based interferometry techniques. We showed in a recent work how a relatively small 10x10 pixel kernel was trained, through machine learning methods, for predicting the locations of phase discontinuities within noisy wrapped phase maps. We describe here how this kernel can be applied in a sliding-window fashion, such that each pixel undergoes 100 phase-discontinuity examinations - one test for each of its possible positions relative to its neighbours within the kernel’s extent. We explore how the resulting predictions can be accumulated, and aggregated through a voting system, and demonstrate that the reliability of this method outp...

A machine learning approach to fringe-location identification F Sawaf and R Groves Delft University of Technology, The Netherlands Machine learning concepts are attracting renewed interest in various computational fields, including machine vision, both in the theoretical and applied domains. Machine learning takes the approach of creating programs which learn how to solve a given problem, rather than programming the machine with exact steps to be performed in order to solve the problem. Before a machine learning system can be deployed, it typically needs to undergo a training phase, which enables it to acquire the necessary structures and information to solve similar problems. The performance of a machine learning system is commonly assessed by measuring how well the system is able to solve problems, which are generally similar but not identical to those used as examples during the training phase. We investigate applying machine learning to a key step of fringe analysis, namely the ...

Phase unwrapping is one of the key steps of interferogram analysis, and its accuracy relies primarily on the correct identification of phase discontinuities. This can be especially challenging for inherently noisy phase fields, such as those produced through shearography and other speckle-based interferometry techniques.
We showed in recent works how a relatively small 10 × 10 pixel kernel was trained, through machine learning methods, for predicting the locations of phase discontinuities within noisy wrapped phase maps. We also demonstrated that our kernel is large enough for effective processing of full-field interferograms. Avoiding, thus, the need for substantially more formidable computational resources which otherwise would have been necessary for training a kernel of a significantly larger size.
In this work, we focus on facilitating a direct comparison between a mathematically rigorous simulation of phase fields versus those obtained in a corresponding real optical setup, both qualitatively and quantitatively. Building on this work is envisage to overcome the issue of generating a large number of training examples whose actual phase discontinuity locations are known a priori with certainty, and go a long way to ensuring that the computer-generated simulations are as representative as possible of the practical optical setup for the target application at hand.

This paper demonstrates that existing Structural Health Monitoring (SHM) techniques have potential during the production phase in addition to their application for maintenance and for in-flight monitoring. Flaws occur during composite fabrication in industry, due to an imperfect process control and human errors. This decreases production efficiency and increases costs. In this paper, the monitoring of Lamb waves in unidirectional carbon fibre (UD-CFRP) prepreg material is demonstrated using both Fibre Bragg Gratings (FBG)s and piezolectric acoustic sensors, and that these SHM sensors may be used for flaw detection and production monitoring. The detection of Lamb waves in a one ply thick sheet of prepreg UD-CFRP material is demonstrated for an FBG sensor aligned with the carbon fibre orientation and bonded to the surface of the prepreg, Furthermore, the velocity of Lamb waves in prepreg UD-CFRP in different orientations is investigated. Finally the successful detection of a material crack in a prepreg UD-CFRP sheet using the Lamb wave detection method is demonstrated.

This report was prepared at the request of the Netherlands Organisation for Scientific Research and the Rijksmuseum Amsterdam with grant support from the Getty Foundation as part of the Panel Paintings Initiative.
ABSTRACT Until the early 17th century almost all portable paintings were created on wood supports, including masterpieces by famous painters, ranging from Giotto to Dürer to Rembrandt. The structural conservation of these paintings requires specific knowledge and skills as the supports are susceptible to damage caused by unstable environmental conditions. Unfortunately, past structural interventions often caused significant damage due to insufficient knowledge of the behaviour of the wood panels, glue and paint layers. Over the last fifty years, the field has developed treatment strategies based on interdisciplinary collaboration and on the knowledge of specialist conservators. Most current conservation protocols rely on empirical knowledge of conservators and are not necessarily based on a scientific understanding of the nature and behaviour of wood and paint layers. In order to move the field forward, it is imperative to strengthen scientific research

This systematic literature review is a part of the Climate4Wood research project of a consortium of Dutch cultural heritage and university partners. The aim of the research project is to identify relative humidity and temperature fluctuations, and related wood moisture changes, that decorated wooden panels can safely sustain. As a consequence, rational guidelines for climate specifications in museums will be developed, in order to balance the cost of energy and the preservation of the collections. This paper provides a concise review of the current knowledge, presented in the literature within the field of climate related damage to decorated wooden panels. The lacunas in the current state of research will give further direction to Climate4Wood. The on-going museum study will be used to visualise the connection between questions posted in the literature study and damage related to historic climate on objects in a museum environment.

We reported in recent works on how a 10x10 pixel kernel can be trained, using machine-learning techniques, for predicting the locations of phase discontinuities within noisy wrapped phase maps. We also showed how such a relatively small kernel can be successfully applied to full-field interferograms, measuring 200x200 pixels or even larger. This was achieved by applying the kernel in a sliding-window fashion, thus increasing its effective neighbourhood size to 19x19 pixels. Boosting is a machine-learning technique which trains relatively weak classifiers in a successive manner, resulting in rapid increases in the quality of predictions produced during the training phase. These improvements have also been shown in the literature to translate into better generalisation to the problem domain, against test examples never encountered during training. We build further on our 10x10 kernel by exploring the potential for applying machine-learning boosting techniques, and report on its merits...

Climate and time are two major factors enhancing the degradation processes of surface and bulk materials. For this reason, estimating the mechanical and chemical deterioration of works of art has been a continuous procedure performed in museums. The developed prototype combines two optical techniques (Fringe Projection and Hyperspectral Imaging in the Visible and SWIR) in one system, enabling the acquisition of the colorimetric, morphologic, chemical and spectroscopic information from the artwork. This combination of: non-destructive techniques; specialized calibration; the use of material, substance, component and pigment databases; data fusion and classification methods, allows the characterization and analysis of canvas and panel paintings. Furthermore, the prototype is portable to allow on-site measurements and analysis. The performance of the prototype is presented from both hardware and software development. Results obtained from real case studies show that mechanical and chem...

The preservation of cultural heritage is a common interest across Europe and the World. In the FP7 Syddarta project the project partners are developing a prototype instrument for the assessment of mechanical and chemical ageing and deterioration on paintings from the Baroque period. Mechanical deterioration is assessed by measuring shape using structured light projection and implementing morphological algorithms. Spectroscopic data is collected by hyperspectral imaging cameras recording in the visible and infrared. This hyperspectral data is processed using Multiplicative Scatter Correction and classified with Principal Com¬ponent Analysis and Support Vector Machines. The processing algorithms for system calibration and assessment of deterioration are described in detail in this paper and examples are presented of the results from painted wooden panels.

Climate and time are two major factors enhancing the degradation processes of surface and bulk materials. For this reason, estimating the mechanical and chemical deterioration of works of art has been a continuous procedure performed in museums. The developed prototype combines two optical techniques (Fringe Projection and Hyperspectral Imaging in the Visible and SWIR) in one system, enabling the acquisition of the colorimetric, morphologic, chemical and spectroscopic information from the artwork. This combination of: non-destructive techniques; specialized calibration; the use of material, substance, component and pigment databases; data fusion and classification methods, allows the characterization and analysis of canvas and panel paintings. Furthermore, the prototype is portable to allow on-site measurements and analysis. The performance of the prototype is presented from both hardware and software development. Results obtained from real case studies show that mechanical and chem...

Phase unwrapping is one of the key steps of interferogram analysis, and its accuracy relies primarily on the correct identification of phase discontinuities. This can be especially challenging for inherently noisy phase fields, such as those produced through shearography and other speckle-based interferometry techniques. We showed in a recent work how a relatively small 10×10 pixel kernel was trained, through machine learning methods, for predicting the locations of phase discontinuities within noisy wrapped phase maps. We describe here how this kernel can be applied in a sliding-window fashion, such that each pixel undergoes 100 phase-discontinuity examinations--one test for each of its possible positions relative to its neighbors within the kernel's extent. We explore how the resulting predictions can be accumulated, and aggregated through a voting system, and demonstrate that the reliability of this method outperforms processing the image by segmenting it into more convention...

This paper discusses a novel application of OCT by using it as a way to study and characterize coatings on wood for a variety of applications and research fields. The coatings studied in this work are transparent layers applied to protect the wood from external influences. OCT can be used to assess the quality or damages of such a coating, to monitor its penetration into the wood structure and to measure its thickness. This study will apply OCT as a non-destructive technique that can both be used to test the quality of applied coatings as well to lay a basis for study coatings on wood in the field of cultural heritage. Measurements of deteriorated coated wood and reference coated wood show that both time- and Fourier-domain OCT are successful in imaging the coated layer. In order to deal with noise and to fully utilize all possible information provided by the obtained data, post-processing plays an important role. Results from the two OCT techniques and their comparison will be presented and discussed.

The FP7 SYDDARTA project had as a main objective of the development of a pre-industrial prototype for diagnosing the deterioration of movable art assets. The device combines two complementary optical techniques, hyperspectral imaging in the visible (400-1000nm) and infrared (1000-2500nm) using electronically tunable optical filters, and 3D scanning using structured light projection. These techniques are integrated in a single prototype allowing the recording of hyperspectral and shape streams, software data fusion and automated estimation of artwork deterioration and degradation using signal processing. The prototype is portable and user-friendly and has been used to study artwork of the baroque period at partner museums in the consortium, IPCHS, Slovenia and RABSAF, Spain. The main challenges posed in the development of such a device were: (i) the integration of reliable and robust illumination units, both for the hyperspectral and 3D channels, which meet the desired requirements of power, homogeneity and stability, (ii) the merging of hyperspectral and 3D data streams in order to construct rich digital representations of the work of art, and (iii) the development of classification strategies based on eigenvector analysis and point cloud segmentation for mechanical alteration and supervised classification by considering spectral samples for the identification and quantification of chemical deterioration.