An official website of the United States government
X-ray fluorescence (XRF) spectroscopy is a common technique in the field of heritage science. However, data processing and data interpretation remain a challenge as they are time consuming and often require a priori knowledge of the composition of the materials present in the analyzed objects. For this reason, we developed an open-source, unsupervised dictionary learning algorithm reducing the complexity of large datasets containing 10s of thousands of spectra and identifying patterns. The algorithm runs in Julia, a programming language that allows for faster data processing compared to Python and R. This approach quickly reduces the number of variables and creates correlated elemental maps, characteristic for pigments containing various elements or for pigment mixtures. This alternative approach creates an overcomplete dictionary which is learned from the input data itself, therefore reducing the a priori user knowledge. The feasibility of this method was first confirmed by applying it to a mock-up board containing various known pigment mixtures. The algorithm was then applied to a macro XRF (MA-XRF) data set obtained on an 18th century Mexican painting, and positively identified smalt (pigment characterized by the co-occurrence of cobalt, arsenic, bismuth, nickel, and potassium), mixtures of vermilion and lead white, and two complex conservation materials/interventions. Moreover, the algorithm identified correlated elements that were not identified using the traditional elemental maps approach without image processing. This approach proved very useful as it yielded the same conclusions as the traditional elemental maps approach followed by elemental maps comparison but with a much faster data processing time. Furthermore, no image processing or user manipulation was required to understand elemental correlation. This open-source, open-access, and thus freely available code running in a platform allowing faster processing and larger data sets represents a useful resource to understand better the pigments and mixtures used in historical paintings and their possible various conservation campaigns.
more »
« less
Can deep learning assist automatic identification of layered pigments from XRF data?
X-ray fluorescence spectroscopy (XRF) plays an important role for elemental analysis in a wide range of scientific fields, especially in cultural heritage. XRF imaging, which uses a raster scan to acquire spectra pixel-wise across artworks, provides the opportunity for spatial analysis of pigment distributions based on their elemental composition. However, conventional XRF-based pigment identification relies on time-consuming elemental mapping facilitated by the interpretation of measured spectra by experts. To reduce the reliance on manual work, recent studies have applied machine learning techniques to cluster similar XRF spectra in data analysis and to identify the most likely pigments. Nevertheless, it is still challenging to implement automatic pigment identification strategies to directly tackle the complex structure of real paintings, e.g. pigment mixtures and layered pigments. In addition, pigment identification based on XRF on a pixel-by-pixel basis remains an obstacle due to the high noise level. Therefore, we developed a deep-learning based pigment identification framework to fully automate the process. In particular, this method offers high sensitivity to the underlying pigments and to the pigments present in low concentrations, therefore enabling robust mapping of pigments based on single-pixel XRF spectra. As case studies, we applied our framework to lab-prepared mock-up paintings and two 19th-century paintings: Paul Gauguin's Poèmes Barbares (1896) that contains layered pigments with an underlying painting, and Paul Cezanne's The Bathers (1899–1904). The pigment identification results demonstrated that our model achieved comparable results to the analysis by elemental mapping, suggesting the generalizability and stability of our model.
more »
« less
- Award ID(s):
- 1743748
- PAR ID:
- 10443685
- Date Published:
- Journal Name:
- Journal of Analytical Atomic Spectrometry
- Volume:
- 37
- Issue:
- 12
- ISSN:
- 0267-9477
- Page Range / eLocation ID:
- 2672 to 2682
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
A longstanding challenge in biology is accurately analyzing images acquired using microscopy. Recently, machine learning (ML) approaches have facilitated detailed quantification of images that were refractile to traditional computation methods. Here, we detail a method for measuring pigments in the complex-mosaic adult Drosophila eye using high-resolution photographs and the pixel classifier ilastik [1]. We compare our results to analyses focused on pigment biochemistry and subjective interpretation, demonstrating general overlap, while highlighting the inverse relationship between accuracy and high-throughput capability of each approach. Notably, no coding experience is necessary for image analysis and pigment quantification. When considering time, resolution, and accuracy, our view is that ML-based image analysis is the preferred method.more » « less
-
Abstract Carotenoid pigments are the basis for much red, orange, and yellow coloration in nature and central to visual signaling. However, as pigment concentration increases, carotenoid signals not only darken and become more saturated but they also redshift; for example, orange pigments can look red at higher concentration. This occurs because light experiences exponential attenuation, and carotenoid‐based signals have spectrally asymmetric reflectance in the visible range. Adding pigment disproportionately affects the high‐absorbance regions of the reflectance spectra, which redshifts the perceived hue. This carotenoid redshift is substantial and perceivable by animal observers. In addition, beyond pigment concentration, anything that increases the path length of light through pigment causes this redshift (including optical nano‐ and microstructures). For example, maleRamphocelustanagers appear redder than females, despite the same population and concentration of carotenoids, due to microstructures that enhance light–pigment interaction. This mechanism of carotenoid redshift has sensory and evolutionary consequences for honest signaling in that structures that redshift carotenoid ornaments may decrease signal honesty. More generally, nearly all colorful signals vary in hue, saturation, and brightness as light–pigment interactions change, due to spectrally asymmetrical reflectance within the visible range of the relevant species. Therefore, the three attributes of color need to be considered together in studies of honest visual signaling.more » « less
-
Flowers have evolved remarkable diversity in petal color, in large part due to pollinator-mediated selection. This diversity arises from specialized metabolic pathways that generate conspicuous pigments. Despite the clear link between flower color and floral pigment production, studies determining predictive relationships between pigmentation and petal color are currently lacking. In this study, we analyze a dataset consisting of hundreds of natural Penstemon hybrids that exhibit variation in flower color, including blue, purple, pink, and red. For each individual hybrid, we measured anthocyanin pigment content and petal spectral reflectance. We found that floral pigment quantities are correlated with hue, chroma, and brightness as calculated from petal spectral reflectance data: hue is related to the relative amounts of delphinidin vs. pelargonidin pigmentation, whereas brightness and chroma are correlated with the total anthocyanin pigmentation. We used a partial least squares regression approach to identify predictive relationships between pigment production and petal reflectance. We find that pigment quantity data provide robust predictions of petal reflectance, confirming a pervasive assumption that differences in pigmentation should predictably influence flower color. Moreover, we find that reflectance data enables accurate inferences of pigment quantities, where the full reflectance spectra provide much more accurate inference of pigment quantities than spectral attributes (brightness, chroma, and hue). Our predictive framework provides readily interpretable model coefficients relating spectral attributes of petal reflectance to underlying pigment quantities. These relationships represent key links between genetic changes affecting anthocyanin production and ecological functions of petal coloration.more » « less
-
Synopsis Flowers have evolved remarkable diversity in petal color, in large part due to pollinator-mediated selection. This diversity arises from specialized metabolic pathways that generate conspicuous pigments. Despite the clear link between flower color and floral pigment production, quantitative models inferring predictive relationships between pigmentation and reflectance spectra have not been reported. In this study, we analyze a dataset consisting of hundreds of natural Penstemon hybrids that exhibit variation in flower color, including blue, purple, pink, and red. For each individual hybrid, we measured anthocyanin pigment content and petal spectral reflectance. We found that floral pigment quantities are correlated with hue, chroma, and brightness as calculated from petal spectral reflectance data: hue is related to the relative amounts of delphinidin vs. pelargonidin pigmentation, whereas brightness and chroma are correlated with the total anthocyanin pigmentation. We used a partial least squares regression approach to identify predictive relationships between pigment production and petal reflectance. We find that pigment quantity data provide robust predictions of petal reflectance, confirming a pervasive assumption that differences in pigmentation should predictably influence flower color. Moreover, we find that reflectance data enables accurate inferences of pigment quantities, where the full reflectance spectra provide much more accurate inference of pigment quantities than spectral attributes (brightness, chroma, and hue). Our predictive framework provides readily interpretable model coefficients relating spectral attributes of petal reflectance to underlying pigment quantities. These relationships represent key links between genetic changes affecting anthocyanin production and the ecological functions of petal coloration.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1039/D2JA00246A
