LAJANUGEN LOGESWARAN : SOLVING JIGSAW PUZZLES USING PATHS AND CYCLES 1 Solving Jigsaw Puzzles using Paths and Cycles

There has been a growing interest in image jigsaw puzzles with square shaped pieces. A solver takes as input square shaped patches of the same size belonging to an image and attempts to reconstruct the image. The key components of a jigsaw solver are a compatibility metric and an assembly algorithm. A compatibility metric uses the color content of the image patches to identify which pairs of pieces are likely to be neighbors in the correct assembly. As the piece size gets smaller, it becomes increasingly difficult for any compatibility metric to confidently identify matching piece pairs. We attempt to exploit more contextual information offered by a compatibility metric compared to previous work to improve its neighbor identification accuracy. We introduce the concept of paths and cycles in jigsaw puzzles and show that they provide a means of identifying correct and incorrect matches. We propose refinement techniques based on this idea which improve the neighbor identification accuracy of a given compatibility metric. We also propose a means of combining the strengths of different compatibility metrics. Whereas recently proposed greedy solvers use the cost values produced by the metric directly to pick piece pairs for assembly, we define a new measure of piece pair compatibility and use it to guide a greedy solver. The proposed solver beats state of the art performance, achieving a mean improvement of more than 15% in absolute placement accuracy for puzzles with piece size 14x14.