Jigsaw Puzzle Solving as a Consistent Labeling Problem

DSpace/Manakin Repository

Show simple item record

dc.contributor.advisor Pelillo, Marcello it_IT
dc.contributor.author Khoroshiltseva, Marina <1979> it_IT
dc.date.accessioned 2022-06-03 it_IT
dc.date.accessioned 2023-02-22T13:45:51Z
dc.date.available 2023-02-22T13:45:51Z
dc.date.issued ND it_IT
dc.identifier.uri http://hdl.handle.net/10579/22748
dc.description.abstract The thesis aims to develop a methodology for solving complex puzzle problems. In the first part of the research, we explore the idea of abstracting the jigsaw puzzle problem as a consistent labeling problem, a classical concept introduced in the 1980s by Hummel and Zucker for which a solid theory and powerful algorithms are available. A formal theory of consistency developed by Hummel and Zucker turned out to have intimate connections with non-cooperative game theory. The theory generalizes classical (boolean) constraint satisfaction problems to scenarios involving ``soft'' compatibility measures and probabilistic (as opposed to ``hard'') label assignments. The problem amounts to maximizing a well-known quadratic function over a probability space which we solve using relaxation labeling algorithms endowed with matrix balancing mechanisms to enforce one-to-one correspondence constraints. The second part addresses the problem of puzzles with eroded borders, a special challenging case of puzzle-solving. Solving puzzles with eroded borders is a common situation when dealing with the re-assembly of archaeological artifacts or ruined frescoes. In this particular condition, the puzzle's pieces do not align perfectly due to the erosion gaps; a direct matching of the patches is consequently unfeasible due to the lack of color and line continuations. To tackle this issue, we propose JiGAN, a GAN-based method for solving puzzles with ruined borders. The experiments on commonly used benchmark datasets demonstrate that our approach is able to address the problem of eroded borders and produce plausible reconstruction results. it_IT
dc.language.iso en it_IT
dc.publisher Università Ca' Foscari Venezia it_IT
dc.rights © Marina Khoroshiltseva, 2022 it_IT
dc.title Jigsaw Puzzle Solving as a Consistent Labeling Problem it_IT
dc.title.alternative it_IT
dc.type Doctoral Thesis it_IT
dc.degree.name Informatica it_IT
dc.degree.level Dottorato it_IT
dc.degree.grantor it_IT
dc.description.academicyear Dottorati_01-07-2022_34-33 Ciclo it_IT
dc.description.cycle 34
dc.degree.coordinator Cortesi, Agostino
dc.location.shelfmark D002196
dc.rights.accessrights OpenAccess it_IT
dc.thesis.matricno 956461 it_IT
dc.format.pagenumber 101 p.
dc.subject.miur INF/01 INFORMATICA it_IT
dc.description.note it_IT
dc.degree.discipline it_IT
dc.contributor.co-advisor it_IT
dc.date.embargoend it_IT
dc.provenance.upload Marina Khoroshiltseva (956461@stud.unive.it), 2022-06-03 it_IT
dc.provenance.plagiarycheck Marcello Pelillo (pelillo@unive.it), 2022-07-01 it_IT

Files in this item

This item appears in the following Collection(s)

Show simple item record