Reassembly Fragments of Images

Reassembly Fragments of Images From 2-pieces approach to N-pieces problem
Maurizio Garbarino
PhD Student

DEI - Seminar Room
September 30th, 2010
16.00

Abstract:

The reassembly of fragments to reconstruct images and objects is a problem often encountered in several scientific fields, such as forensics, medicine and archaeology.
In the field of archaeology, it is very common to find painting fragments during excavation. The manual execution of the reassembly task can be very difficult, as it requires great amount of time, skill and effort. Thus, a method for the automation of such a work is fundamental and can significantly reduce the human effort involved leading to faster, and more efficient painting reassembly.

The starting point of the presented project is a method for 2-pieces matching that uses only information extracted from the outlines and from the color contents of the fragments, without relying on any knowledge of the final image.
The reassembly of multiple pieces is formulated as a combinatorial optimization problem and image assembly is then done by finding an optimal ordering of fragments matching. Various algorithms are analyzed and a comparison of speed and accuracy is reported. Finally, is discussed whether the use of a 2-pieces matching algorithm as base for N-pieces reassembly is good and fast enough to be used in real applications.

Contacts:
Maurizio Garbarino

Research Area:
Artificial intelligence, robotics and computer vision