Genetic Algorithm with Wisdom of Artificial Crowds Solution for Assembling Jigsaw Puzzles

In the square jigsaw puzzle problem, a complete image is supposed to be reconstructed from a set of non- overlapping, unordered pieces. To fulfill this objective, we have proposed a hybrid approach of Genetic Algorithm and Wisdom of Artificial Crowds. The objective function considers only the information of the pixel value on the borderline of the piece. The main contribution of the algorithm is focused on using the concept of WOC to improve the solutions provided by GA where data on border line of the pieces were used in the puzzle assembly. By introducing a novel cross-over technique, we were able to arrive at better expert individuals. A new best neighbor greedy approach was also proposed for reconstruction phase of the puzzle. A strong WOC effect was observed where the average performance of the aggregation method outperformed even the best individual in most of the cases especially where we have large values for the puzzle size.