Studying Digital Imagery of Ancient Paintings by Mixtures of Stochastic Models

• Activity Number: 281
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2003 : 2:00 PM to 3:50 PM
• Sponsor: IMS
• Author(s): Jia Li*+ and James Wang
• Companies: Pennsylvania State University and Pennsylvania State University
• Keywords: mixture of stochastic models ; 2-D multiresolution hidden Markov model ; image classification and clustering ; digital imagery of paintings

Abstract

This paper addresses learning based characterization of fine art painting styles. The research has the potential to provide a powerful tool to art historians for studying connections among artists or periods in the art history. Depending on specific applications, paintings can be categorized in different ways. We focus on comparisons among artists. To profile the style of an artist, a mixture of stochastic models, in particular, the 2-D multiresolution hidden Markov model (MHMM), is estimated using training images, forming a digital signature of the artist. For certain types of paintings, for instance, numerous Chinese paintings in ink without engaging color or even tone, strokes provide reliable information to distinguish artists. The 2-D MHMM allows us to analyze images at a relatively global level and hence more likely to capture properties of the painting strokes. The mixtures of 2-D MHMMs established for artists can be further used to classify paintings and compare paintings or artists. We implemented and tested the system using high-resolution digital photographs of some significant Chinese paintings in history.