Fast Placement Algorithm for Rectilinear Jigsaw Puzzles

Abstract

In previous papers, rectilinear jigsaw puzzles have been described as a specialized placement problem such that this has at least one solution to placement, but generally not that many. Instead of adopting the well-known iterative method to solve this problem, a new game-theoretic algorithm is developed by translating the problem to one of checkmate in games analogous to chess or shogi. By extending the game-theoretic algorithm, a much faster algorithm for large puzzles is developed by introducing various heuristics on the placement of pieces. We also proved through numerical experiments that this worked efficiently.