A multi-spectral camera was developed by the CRISATEL consortium in a EU IST project 2001-2005. Hundred paintings conserved in several museums were scanned by the C2RMF. It produces 13 very high definition (12.000x20.000 pixels) image files after filtering of the reflected light in short bandwidths (40 nanometres) to cover the visible spectrum and the near infra-red (from 380 to 1050 nm). After the corrections of the camera electronic noise, the CCD sensitivity, the geometric size of the 13 images and the topographic heterogeneity of the lighting, we can simulate a color image of the painting for any lighting source by means of the spectral reflectance curves of each pixel. With these experimental conditions, we are no more dependant of the lighting source quality as it is with a RGB camera.

Why to use image processing to study canvas paintings in a museum laboratory ? Because the computer screen visualisation of large size image made at different wavelengths, allows us to locate information at different level of depth in the paint layer. Near the ultra-violet and the blue, the image correspond to the reflectance of the upper layers of the painting (the varnish firs). Close to the red and in the infra-red, the paint layer is partly transparent and we are able to see the under-painting such as under-drawing. The development of tools to examine by computer the same magnified area into the paint layer is a new mode for interpretation of documents using false color imaging or semi-transparency of couple of images.


As we write in the GPU Spectral Viewer section our goal here is to provide a specific image processing tool for art preservation and analysis using reflectance images. In this section we will describe all the basic colorimetric computation used in GPU Spectral Viewer.

Color is the perceptual result of light in the visible region of the spectrum, having wavelengths in the region of 380 nm to 780 nm. The human retina has three types of color photoreceptor cells cone, which respond to incident radiation with somewhat different spectral response curves. Because there are exactly three types of color photoreceptor, three numerical components are necessary and theoretically sufficient to describe a color.

The CIE color standard is based on imaginary primary colors XYZ i.e. which don't exist physically. These virtual primary colors have, however, been selected so that all colors which can be perceived by the human eye lie within this colour space. The XYZ system is based on the response curves of the three color receptors of the eye's. Since these differ slightly from one person to another person, CIE has defined a "standard observer" whose spectral response corresponds more or less to the average response of the population. This objectifies the colorimetric determination of colors.

In our case we have to determine for each pixel of the multispectral image the resulting XYZ value generated by the virtual light interaction with the reflectance image using the following formulae:


Where R() is the reflectance spectrum and L() is the light spectrum (used as illuminant). In this application we introduced an other feature allowing varnish removal using the transmittance properties of the varnish. The CRISATEL project produce 13 planes images which correspond to the following frequencies: 400, 440, 480, 520, 560, 600, 640, 680, 720, 760, 800, 900 and 1000nm. Only the 10 first planes interact with the visible part of the light. Considering this, the previous formulae become:


Using this formulae we can compute the resulting XYZ values for each pixel of the source image. Of course, the obtained values are only an approximation of the values that will be measured in front of the original painting.

Top