Last edited on 2005-01-17 15:05:54 by stolfi

Fragment Re-Assembly Project - Images

Test with flat fragments

Helena Leitão's doctoral thesis project was to build a software tool to spped up the reassembly of fragmented ceramic objects.

Due to equipment limitations, that project considered only flat ceramic objects, whose fragments could be digitalized on a tabletop document scammer. The software was therefore tested with 111 "fresh" fragments, which we created by breaking 5 unglazed ceramic tiles, each about 5 × 20 cm.

• The test fragments, manually reassembled in their original positions, imaged on a tabletop scanner:
  [small BIG]
• The test fragments in random positions, imaged with a tabletop scanner. These images are the input to our software:
  [small BIG]
• Fragment outlines automatically extracted from the images above:
  [small BIG]
• Outlines of fragment pairs that were adjacent before fracture, in their correct relative positions, showing the precision of fit. Corresponding parts of the outlines are highlighted in red and blue:
  [small BIG]
• Some of the pairs of fragments with similar outlines that were found by Helena's programs, soreted by decreasing degree of similarity. Pairs marked with a green spot are correct (i.e. the fragment were indeed adjacent in the original objects). Unmarked pairs are false positives; that is, pairs which were not adjacent in the original object, but were considered similar by the program:
  [small BIG]
• Virtual reconstruction of the fragmented objects. The parts highlighted in red are pairs of outline segments that were found to be similar by the program. Orange lines connect the numbers of the corresponding fragments:
  [small BIG]

Images of ral archaeological fragments

• A few fragments of Brazilian Colonial-era ceramics, unearthed from under the Imperial Palace in Rio de Janeiro and stored at the National Museum (MN) at Quinta da Boa Vista:
  [small BIG] [small BIG]
• Two adjacent Colonial-era ceramic fragments from the MN holdings. Note the precision of fit:
  [small BIG]
• Archaeological fragment of indian pottery recovered from the Caju site (Campos, RJ; IX century AD) by the Brazilian Archaeology Institute (IAB), after washing, awaiting to be numbered:
  [small BIG] [small BIG]
• Close-ip image of a fragment from Caju, with the lot number:
  [small BIG]
• Some of the ceramics fragments stored at the IAB deposit, in Belfort Roxo, RJ:
  [small BIG]
• Close-up images of funerary urns and fragments thereof, matched during recovery. These pictures show how well adjacent fragments fit together even after eleven centuries in the ground:
  [small BIG] [small BIG] -- [small BIG] -- [small BIG]