| Abstract: | SUMMARIES. An experiment was performed that compared conventional small-aperture and image-based reflection spectrophotometry of paintings. The imaging system used a liquid-crystal tunable filter, resulting in 31 spectral bands evenly sampled between 400 and 700 nm and ranging in bandwidth betweeti 10 and 60 nm. The small-aperture spectrophotometer had a constant bandwidth of 10 nm. Test targets consisting of chromatic and neutral samples of various colors and spectral properties were used to derive a calibration trans¬formation between the two technologies. Three paintings were analyzed: Saint Jerome Reading by Alvise Vivarini, Murnau by Alexej von fawlensky and Pot of Geraniums by Henri Matisse, all from the collection of the National Gallery of Art, Washington, DC. Average colorimetric accuracy varied between 2.0 and 3.2 ∆E00 units and the average spectral accuracy varied between 1.0 and 2.1% spectral root-mean-square. Two drawbacks are that the imaging system has a high uncertainty at short wavelengths, and the spectral matches for samples with flat spectra are slightly worse than for other samples. Both limitations can be corrected by changes in lighting, the calibration target, and the method of deriving the transformation matrix. Nevertheless, the imaging system has the advantage of no moving parts and may not require image registration, making it well suited to perform scientific imaging of cultural heritage. Furthermore, the image-based spectra have sufficient accuracy for pigment identification and mapping.
CONCLUSION. Image-based spectrophotometry was compared with traditional, small-aperture spectrophotometry by analyzing three paintings that have a range of coloration, spectral properties, and surface attributes. The correla¬tion between the two methods of spectral measurement was reasonable, resulting in average colorimetric differ¬ences between 2.0 and 3.2 AE„„ color difference units and average spectral RMS differences between 1.0 and 2.1%. There were systematic differences caused by the poor signal-to-noise properties of the imaging system at short wavelengths and interrelationships between measurement geometry and surface attributes. There are several opportunities for improvement. First, the calibration targets had a very small range of reflectances at the shortest wavelengths, caused by the use of titanium dioxide white. It would be interesting to replace the titanium white with a different scattering pigment that could yield higher reflectance at shorter wavelengths. This would improve accuracy for short wavelengths when imaging paintings containing lead white. The poor signal-to-noise properties at short wavelengths were caused by the detector, the LCTF and the light source having low sensitivity, transmitían ce and radiance, respectively. An obvious remedy would be to replace the tungsten-halogen lights by a source with greater short-wavelength radiance such as a Xenon lamp. The transformation matrix treated each wavelength independently. As a consequence, spectra with regions of constant reflectance factor, such as neutrals, had excessive spectral variability. Adding a smoothness constraint to the matrix could improve performance. An alternative approach would be to add a weighting function to each sample comprising the calibration target; weighting the neutral samples more heavily would improve smoothness. This could also provide opportunities for improved performance for specific colorants. The transformation could be optimized to achieve best estimation accuracy for certain colors. The targets used in this experiment were not designed for direct imaging of cultural heritage. It seems likely that improvements in target design will improve spectral estimation accuracy; this is an active area of research for the authors [33, 43]. The transformation matrix was optimized to minimize spectral RMS error. This does not lead to minimum color difference. Further optimization could be performed to improve colorimetric accuracy for a specific illuminant and observer, an approach used by the authors when imaging with a RGB digital camera [25-27]. Although the system can be improved, it is nonethe¬less well suited as an analytical spectral instrument. By adding a computer-controlled lens-focusing system, image acquisition can be fully automated. Since the spectral sampling is computer controlled, there are many opportunities to improve the data by having uneven spectral sampling, reducing the number of channels, and adding wide-band acquisition by temporal processing in addition to the usual spectral processing. |
|---|
