A solution to the Maze problem with Dijkstra

The general idea:
1) Think of every pixel in the maze as node on a connected graph
2) Define the walls as having high weights. This ensures that walls will act as separators
3) Use a 4-connected neighborhood to link adjacent pixels/nodes
4) Convert the maze image into a sparse Distance matrix(similar to an adjacency matrix with weights instead of ones).
5) Use graphshortestpath() from the bioinformatics toolbox to find the shortest path