Estimating the Location of Illuminants in Realist Master Paintings
International Conference on Pattern Recognition, Hong Kong, 2006
Abstract
We address the problem of estimating the position of illuminants in realist master paintings. To this end, we apply an algorithm from computer vision that has been used to detect forgeries and composites in digital photographs. Our contributions are three-fold: (1) we introduce the application of a two-dimensional, model-free occluding contour algorithm to the problem of locating the source of illumination in realist master paintings, (2) we introduce a maximum-likelihood criterion and computational method for integrating estimates from multiple occluding contours, and (3) we apply our methods to a key painting adduced as evidence in a recent controversial theory about painting praxis in the Renaissance and Baroque.