Denoising of underdrawing IR images by means of TPS

Infrared analysis has long been used in the fields of art history and restoration to determine the authenticity of artwork, to establish the period of an anonymous work, and above all to unmask underdrawings. An underdrawing is a preparatory drawing for a painting, sketched directly on a ground. Underdrawings are typically sketched using charcoal, but artists have been known to also use chalk, pencil or paint and brush and other media. The underdrawings are later covered with the artist’s medium, and the underdrawings are not visible any more. Infrared analysis provides an easy nondestructive mean of eliminating the overlying paint on the underdrawings of many artworks.


Images acquired with infrared camera appear grayish (and may therefore have poor contrast) and large amount of noise is often present. The following analysis of infrared images of artworks is much easier in case of denoised images. The proposed denoising method is based on the thin-plate splines. The new weighting function has been introduced, which diversifies the influence of individual pixels for computation of approximating surfaces. This approach enables to decrease unwanted smoothing effect of TPS denoising on edges and small image details while achieving even bigger noise removal than the original method.

Example of the noisy IR data (left) and denoised version (right) using the new method.



 and small details from the previous pictures (noisy - left, denoised - right)


Duration: 2000-2002
Contact person: Barbara Zitova
Involved etern: Roman Kašpar 



  • Kašpar Roman, Zitová Barbara: Weighted thin-plate spline image denoising, Pattern Recognition vol.36, 12 (2003), p. 3027-3030
  • Kašpar Roman, Zitová Barbara: Weighted thin-plate spline image denoising, Computer Analysis of Images and Patterns. Proceedings, p. 730-737 , Eds: Petkov N., Westenberg M. A., Springer, (Berlin 2003), International Conference CAIP 2003 /10./, (Groningen, NL, 25.08.2003-27.08.2003)