2D fragmented object reconstruction with the use of the chromatic and thematic content

In this article, we examine the reconstruction of seventeenth century BC fragmented wall paintings, excavated at Akrotiri, Santorini, Greece. We use a previously presented algorithm for obtaining potential fragment matches based on the shape of the fragments. We then extract additional information from the fragments’ images in order to drastically reduce the number of potential matches proposed by this system. In particular, we examine the performance of several criteria that make use of the chromatic (color) content of the fragments’ images near the edge of the piece. We also extract additional information regarding the thematic content of the fragments; this information is then used to establish an additional criterion, which pertains to the continuity of the thematic content across the segments of the fragments’ edge border that have been found to match with regard to shape. We first apply these criteria on the reconstruction of a commercial jigsaw puzzle in order to evaluate their performance. We then proceed to apply the criteria on several images of the Santorini fragments. For reasons of comparison, we also apply the color matching algorithms introduced by other researchers. Results show that in the presence of color deterioration and noise introduced by the passage of time and the fragmentation process, the combination of the information extracted from both the chromatic and the thematic content of the fragment images yields clearly superior results by those obtained from the use of chromatic information alone.