Virtual recomposition of frescos: separating fragments from the background

The paper addresses the segmentation of fragments from the background using a clustering approach in the color space. The problem is involved in the on-going development of a digital system for the virtual aided recomposition of fragmented frescos, an innovative approach currently being proved on the S. Matthew's fresco, made by Cimabue for the Upper Church of S. Francis in Assisi and broken, during the earthquake in 1997, into more than 140,000 pieces. The developed technique has been used to build the fragments' database. The segmentation process is based on the fast global k-means algorithm. Experiments show that separation of foreground and background on a so large number of images is much more manageable and effective using prototypes built by the clustering algorithm.

[1]  Nikos A. Vlassis,et al.  The global k-means clustering algorithm , 2003, Pattern Recognit..

[2]  Jing Li Wang,et al.  Color image segmentation: advances and prospects , 2001, Pattern Recognit..