The Information Initiative at Duke is merging mathematical sciences and the arts—with beautiful results. Within the realm of visual arts, faculty and students are working with museums, artists, and historians to illuminate authenticity—and forgery—and to preserve history. In the process, we are transforming what is meant by liberal arts education.
James B. Duke Professor Ingrid Daubechies is one of the country’s most prominent mathematicians, and the first woman president of the International Mathematical Union. She is world renowned for her pioneering contributions to the theory and applications of wavelets and filterbanks. The results of her work are used daily in millions of high-tech products.
Daubechies also uses math to compare works of art and identify forgeries. She turns a painting into a massive stream of data using wavelet transforms and other image analysis tools. Daubechies’ system “crunches” a painting into data by slicing it into multidimensional patches and scanning for patterns of color and other features. This deep, granular look reveals the artist’s unique style and spontaneity within brushstrokes. It also detects overly careful, impostor work.
“If you try to make a copy, you pay so much attention to what you’re doing that you probably paint more slowly and with a more restrained hand,” says Daubechies. The difference may not be detectible by the naked eye, but it’s quantifiable.
Daubechies is currently working with the North Carolina Museum of Art (NCMA) and Dutch artist and art historian Charlotte Caspers on another project—the construction of a missing panel of an altarpiece by Francescuccio di Cecco Ghissi, a 14th century Italian painter. Daubechies and her collaborators are using math to “age” their replacement panel by identifying crack patterns, color degradation, and other details in the other panels, and then recreating these elements.
Daubechies’ students are getting in on the action, putting their analytical and engineering skills to use by designing specialized approaches. Third-year graduate student Rachel Yin has put together
a software package that digitally removes artifacts on X-rays of paintings on wood panels that were “cradled” by conservators in the early 20th century. The cradling is a hardwood reinforcement attached to the back of the thinned wood panels. They appear as a dominant feature on X-rays, which obscures their “readings” by art conservators.
Yin presented these results at the prestigious International Conference on Image Processing held in Paris in October 2014. Daubechies and the NCMA have applied for a grant that would allow the translation of this package into a user-friendly Photoshop plugin that would be free for art conserva- tors and art historians to use via the internet.
Undergraduates are involved in exciting ways as well. They are working on NCMA’s Ghissi panels and are distinguishing copies from original artworks for independent study and other projects.