Francisco H. Imai

Perceptual color difference metric for complex images based on Mahalanobis distance

In this paper, a perceptual color difference is presented as an alternative color difference metric for complex images instead of the conventional color difference equations. This color difference is derived based on Mahalanobis distance by using covariance matrices for differences of each color attribute. The covariance matrices for each class of images can be obtained by psychophysical experiments using just noticeable difference in paired comparisons. We compared the resultant matrices for different class of images and the information in the matrix can give very useful trends and clues about which kind of transformation can minimize the perceptual color difference in images when a transformation such as gamut mapping is required.

Multispectral-based color reproduction research at the Munsell Color Science Laboratory

The traditional techniques of image capture, scanning, proofing, and separating do not take advantage of colorimetry and spectrophotometry. For critical color-matching applications such as catalog sales, art-book reproductions, and computer-aided design, typical images, although pleasing, are unacceptable with respect to color accuracy. The limitations that lead to these errors have a well-defined theoretical basis and are a result of current hardware and software. This has led us to a re-examination of the traditional graphic reproduction paradigm. A research and development program has begun that will alleviate the theoretical limitations associated with traditional techniques. There are four main phases: (1) Multi-spectral image capture, (2) Spectral-based separation and printing algorithm development, (3) Implementation on press, and (4) Systems integration with data and image archives. This paper describes this new paradigm, summarizes recent research results, and considers implementation opportunities.

Spectral reproduction from scene to hardcopy: I. Input and output

Efforts to construct end-to-end color reproduction systems based on the preservation of scene spectral data have been underway at the Munsell Color Science Laboratory. The goal is to present hardcopy results which are spectrally matched to original colors. The evaluated approach consists of capturing scenes through a trichromatic digital camera combined with multiple filterings followed by an image processing stage and then four-color printing. The acquisition end is designed to estimate original scene spectra on a pixel-by-pixel basis based on system characteristics which takes into account the camera sensitivities as modulated by the filterings followed by an image processing stage and then four-color printing. The acquisition end is designed to estimate original scene spectra on a pixel-by-pixel basis based on system characterizations which takes into account the camera sensitivities as modulated by the filterings an scene colorant make-up. The spectral-based printing used in this research is able to produce the least metameric reproduction to the original scene using a computationally feasible approach. Results show a system accuracy of mean (Delta) E*94 of 1.5 and spectral reflectance rms error of 0.9 percent.

A Comparison of Small-Aperture and Image-Based Spectrophotometry of Paintings

Abstract 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 between 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 transformation between the two technologies. Three paintings were analyzed: Saint Jerome Reading by Alvise Vivarini, Murnau by Alexej von Jawlensky 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 ΔE00 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.

Spectral estimation of artist oil paints using multi-filter trichromatic imaging

A practical and easy way to capture images of oil-paintings and estimate their spectral reflectance as a function of position was tested. For the image acquisition, a trichromatic digital camera was used in conjunction with an absorption filter producing six channels. From an a priori statistical analysis of common artist oil paints, spectral reflectance was estimated. These experiments showed that it is possible to estimate the spectral reflectance with an accuracy of average ΔE*94 of 1.7 and spectral reflectance rms error of 2.2%. Of particular interest is guidance towards the design of a universal calibration target for imaging paintings.

Spectral reproduction from scene to hardcopy: II. Image processing

Traditional image processing techniques used for 3- and 4- band images are not suited to the many-band character of spectral images. A sparse multi-dimensional lookup table with inter-node interpolation is a typical image processing technique used for applying either a known model or an empirically derived mapping to an image. Such an approach for spectral images becomes problematic because input dimensionality of lookup tables is proportional to the number of source image bands and the size of lookup table sis exponentially related to the number of input dimensions. While an RGB or CMY source images would require a 3D lookup table, a 31-band spectral image would need a 31-dimensional lookup table. A 31-dimensional lookup table would be absurdly large. A novel approach to spectral image processing is explored. This approach combines a low-cost spectral analysis followed by application of one from a set of low-dimensional lookup tables. The method is computationally feasible and does not make excessive demands on disk space or run-time memory.

Computer Vision, Imaging and Computer Graphics - Theory and Applications : International Joint Conference, VISIGRAPP 2015, Berlin, Germany, March 11-14, 2015, Revised Selected Papers

Computer Vision, Imaging and Computer Graphics - Theory and Applications : International Joint Conference, VISIGRAPP 2015, Berlin, Germany, March 11-14, 2015, Revised Selected Papers

Front Matter: Volume 8299

This PDF file contains the front matter associated with SPIE Proceedings Volume 8299, including the Title Page, Copyright Information, Table of Contents, and the Conference Committee listing.