Improvement of the digital radiographic images of old paintings on wooden support through the anisotropic diffusion method

Abstract

[EN] The main defect types of historic-artistic paintings on wood are ruptures, scratches, twisting and suchwhich may be inflicted by environment conditions, insects, dust, and dirt as well as by physical damage.The exact localization of the defects and determination of their extent may be achieved using industrialradiography as a non-destructive testing method. The radiographs thus produced may suffer from blurri-ness mainly due to the inherent scattering of X-rays especially in the case of paintings on a wooden baseand hindering therefore accurate detection of the size and shape of such defects. Image processing meth-ods have been employed to reduce the blurriness of images leading to improved analysis of the images. Inthis study, an image processing method based on anisotropic diffusion with an automatic threshold levelwas applied to achieve improved outcomes. The reconstructed images of the implemented algorithmyielded sharper edges. Defects such as those due to xylophagous attack, the effect of the brushstrokes,superficial fissures, oxidation of the nails, and the different types of construction woods were bettervisualized than from the original image. The algorithm was shown to be useful by operators includingpainting conservators for their procedures.