9 minute read
Two Is Better than One
from OCT 2022
True Color Museum Imaging Using Dual Illumination
By Olivia R. Kuzio and Susan P. Farnand, Rochester Institute of Technology; Rochester, NY/US
Spectral imaging has long been recognized to outperform conventional color imaging for color accurate reproduction of cultural heritage objects and materials.(1-4) Many such materials are color inconstant, meaning that, when the lighting changes, their appearance changes, sometimes dramatically. This happens because seeing color and capturing color are not identical processes. Keeping everything else exactly the same, if you substitute your eyes for a conventional color camera, the picture you take will not exactly replicate what you see – your eyes and conventional cameras are not equivalent sensors. There are standard methods of transforming RGB camera signals to trichromatic human color perception, called profiling. Creating an objective camera profile is essentially the process of determining a mathematical map between these two sensors that transforms camera capture to match human vision as closely as possible, but problems remain in ‘as closely as possible’. When working to create a master file, from which digital and print reproductions can be tuned for specific viewing and illuminating conditions, ‘as closely as possible’ may not be good enough. Spectral imaging, in comparison, involves sampling the color spectrum more finely than conventional color capture.(5-6) However, the cost and complexity of the equipment and workflows for carrying out this kind of imaging have precluded its widespread adoption by heritage imaging professionals for routine imaging and digitization projects.
Past efforts to address these limitations have made progress toward both image capture and processing strategies that simplify the implementation of spectral imaging for cultural heritage applications.(7-9)
The research described here was motivated by a desire to build upon this foundation and further lower the perceived barriers-to-entry around spectral imaging. Specifically, this was realized through the development of a user-friendly software application for processing spectral image sets captured using an RGB camera + dual-illumination strategy that is practical for implementation within a studio photography environment.(9-11)
Dual illumination
What do we mean by ‘dual illumination’? The software application that has been developed, called Beyond RGB(12), takes as input two RGB images captured under two different lighting conditions. By changing the lighting conditions, we effectively change the way the camera ‘sees’ the object, and because of this, the red, green, and blue channels of each RGB image end up containing slightly different information. Combining them together results in a stack of six unique channels (Figure 1A). This ‘dual-RGB’ spectral imaging technique, which uses optimized LED illumination allows us to efficiently use familiar cameras as spectral imagers (Figure 1B).
Beyond RGB
Spectral capture strategies that utilize professional-level consumer cameras typically found in cultural heritage imaging studios have been developed and refined over the past two decades.(4,13-15) At RIT, software tools have also been developed alongside these workflows for processing and viewing the resulting spectral image sets.(16-18) While these software tools are adaptable and effective within the research settings in which they were developed, more robust and intuitive processing solutions may encourage experimentation with and eventually more routine use of spectral imaging and archiving by a wider group of practitioners. Recognizing this, a team of RIT software engineering students was recruited to create a software application that would facilitate processing of spectral image master files captured using the dual-illumination spectral capture technique.(11)
Beyond RGB was designed with user experience in mind, combining technical utility with a graphical user interface that is simple and intuitive. Using pairs of RGB images captured under the dual illumination conditions as the inputs, it performs colorimetric and spectral calibrations, and outputs a color calibrated RGB image, a spectral reflectance calibration that enables interactive material reflectance estimation, and supplementary data files that document the details and accuracy of both calibrations. An sRGB preview of the final color calibrated image that can be zoomed to show pixellevel detail can be viewed through the application’s Image Viewer functionality (Figure 2). Additionally, the built-in Spectral Picker allows the user to select regions of interest from which to display and export estimated reflectance spectra.
While the main focus of the application is encouraging the capture of spectral master files and enabling the calibration and export of the color managed RGB image, the ability to perform spectral estimation may be of more interest in future versions of the software that expand upon the more familiar applications of spectral imaging, like pigment identification and mapping.
Preliminary Results
Beyond RGB was tested with images from a variety of cameras and targets representative of those used in photography studio settings. Typical colorimetric and spectral calibrations based on these images resulted in less than about 2 ∆E-00 and 3 ∆E-00 mean color difference and less than about 10% and 20% maximum spectral reflectance difference for calibration targets and verification targets, respectively. These results are of course highly equipment- and setup-dependent; details of the typical kit and setup used to capture images for testing can be found in.(10)
With the initial version of Beyond RGB in hand, we recruited a class of museum studies students to test the software on a variety of computers, including their personal laptops, to get a feel for the ease of distributing, setting up, and running for novice users. Additionally, we traveled to work with several cooperating institutions who generously offered to host us to demonstrate and utilize the software using a portable spectral imaging, education, and training kit (Figure 3).
Conclusions and Future Work
Beyond RGB is a spectral image processing software that has been developed to promote capturing and archiving spectral master images in photography studio settings. Release v1.0.0 is a living, updatable, open-source project, and is freely available for download from the project’s public GitHub repository. It is a fully functional build, but it is the first public release of the software, and it should be noted that the timeline for foundational build did not allow for rigorous user testing. Currently, a new team of RIT senior software engineers is diving into the process of developing and implementing improvements and feature additions in Beyond RGB v2.0. This list already includes several features that will improve the functionality of Beyond RGB, including batch processing, a more comprehensive User Guide, and much more.
The development and distribution of Beyond RGB represents the next step in continued efforts to create more practical methods for integrating spectral capture strategies with cultural heritage imaging and archiving. We hope that our targeted users will be inspired to experiment with the software in their own environments, and to communicate their experience and feedback to us as we endeavor to build upon this work. ■
Author Biographies
Olivia Kuzio is a PhD candidate in the Program of Color Science at Rochester Institute of Technology, where she conducts research in the Studio for Scientific Imaging and Archiving of Cultural Heritage. During her graduate studies, she has completed conservation science internships at the Smithsonian Museum Conservation Institute and the Getty Conservation Institute. She holds BS and MS degrees in chemistry from Pennsylvania State University and the Rochester Institute of Technology, respectively.
Susan Farnand is an Assistant Professor at Rochester Institute of Technology. Her main research areas center around human color vision and perception and include visual attention, color imaging, image quality metrics, 3D printing, and archiving. Prior to joining RIT in 2006, Dr. Farnand was a senior research scientist at Eastman Kodak Co. working primarily on projects in perceptual image quality measurement and modeling. She holds a BS in engineering from Cornell University, and an MS in imaging science and PhD in color science from RIT. Dr. Farnand is IS&T President.
References
1. Saunders, D.; Cupitt, J. Image Processing at the National Gallery: The VASARI Project. Natl. Gall. Tech. Bull. 1993, 14, 72–85.
2. Martinez, K.; Cupitt, J.; Saunders, D.R. High-Resolution Colorimetric Imaging of Paintings. In Proceedings of the Cameras, Scanners, and Image Acquisition Systems; May 20 1993; Vol. 1901, pp. 25–36.
3. Ribés, A.; Brettel, H.; Schmitt, F.; Liang, H.; Cupitt, J.; Saunders, D. Color and Multispectral Imaging with the CRISATEL Multispectral System. In Proceedings of the PICS’03 The Digital Photography Conference; 2003; pp. 215–219.
4. Berns, R.S.; Taplin, L.A.; Nezamabadi, M.; Mohammadi, M.; Zhao, Y. Spectral Imaging Using a Commercial Colour- Filter Array Digital Camera. In Proceedings of the The 14th Triennial ICOM-CC Meeting; 2005; pp. 743–750.
5. Staniforth, S. Retouching and Colour Matching: The Restorer and Metamerism. Stud. Conserv. 1985, 30, 101–111.
6. Berns, R.S. Color-Accurate Image Archives Using Spectral Imaging. In Scientific Examination of Art: Modern Techniques in Conservation and Analysis; Washington, DC, 2005; pp. 105–119.
7. Kuzio, O.R.; Berns, R.S. Color and Material Appearance Imaging and Archiving Using a Sony Alpha A7R III Camera; Rochester Institute of Technology, 2018;
8. Kuzio, O.; Farnand, S. Color Accuracy-Guided Data Reduction for Practical LED-Based Multispectral Imaging. In Proceedings of the Archiving 2021 Final Program and Proceedings; 2021; pp. 65–70.
9. Kuzio, O.; Farnand, S. LED-Based versus Filter-Based Multispectral Imaging Methods for Museum Studio Photography. In Proceedings of the Proceedings of the International Colour Association Conference 2021; 2021; pp. 639–644.
10. Kuzio, O.; Farnand, S. Comparing Practical Spectral Imaging Methods for Cultural Heritage Studio Photography. J. Comput. Cult. Herit. 2022, in press.
11. Kuzio, O.R.; Farnand, S.P. Beyond RGB: A Spectral Image Processing Software Application for Cultural Heritage Studio Photography. In Proceedings of the Archiving 2022 Final Program and Proceedings; 2022; pp. 95–100.
12. Dalesio, P.; Hammerstone, A.; Knox, T.; O’Neil, J.; Ponzetti, J. Beyond RGB, 2022, https://github.com/tjdcs /Imaging-Art-beyond-RGB.
13. Shrestha, R.; Hardeberg, J.H. Multispectral Imaging Using LED Illumination and an RGB Camera. In Proceedings of the 21st Color and Imaging Conference Final Program and Proceedings; 2013; pp. 8–13.
14. Berns, R.S. Theory and Practice of Dual-RGB Imaging; Rochester Institute of Technology, 2016;
15. Berns, R.S. Digital Color Reconstructions of Cultural Heritage Using Color-Managed Imaging and Small-Aperture Spectrophotometry. Color Res. Appl. 2019, 44, 531–546.
16. Cupitt, J.; Martinez, K. VIPS: An Image Processing System for Large Images. In Proceedings of the Proceedings of SPIE 2663, Very High Resolution and Quality Imaging; 1996; pp. 19–28.
17. Berns, R.S.; Chen, T. Update:* Practical Total Appearance Imaging of Paintings. In Proceedings of the IS&T Archiving Conference Proceedings; 2012; pp. 162–167.
18. Studio for Scientific Imaging and Archiving of Cultural Heritage: Software Available online: https://www.rit.edu/ science/studio-scientific-imaging-and-archiving-cultural-heritage#software (accessed on 17 March 2022).