Due to sophisticated walking characteristics of pedestrians, utilizing the street network is not an optimum solution to calculate the route for pedestrians. Mostly in all the navigators for pedestrian, route calculation system uses the street network without paying attention to one-way roads which means that the pedestrian can walk through both directions. In fact pedestrians like to make shortcut through the parking lots, buildings, parks and stairs but their decision almost is not the shortest path as they are not aware of the distances. If we classify the satellite image to white and black pixels (Picture elements) or binary images, assuming that the white is allowed and the black is forbidden for the pedestrian to walk or classify the ortho-photo based on walk-able and non-walk-able area, it is possible to define an algorithm finding the shortest path from origin to the target via artificial intelligence methods. If we define zero for the position of the target and start to increase the adjacent positions (up, down, left and right) by one and do the same until the position of origin gets a value such as n, and then start from the origin to move back toward the target by means of positions of n-1, n-2, ?, 0 then the result would be the shortest path. In the algorithm the pedestrian doesn?t have right to go through obstacles. The position of the obstacles or non-walk-able area can be defined as a value such as -1. We do not need to change the value of the positions of obstacles while calculating the route as shortest path. An array can be assigned to the classified photo and these values can be changed until, the origin receives a value which means that the route already calculated and shortest path found or the pointer of the recursion goes back to the target which means that there is no rout from the origin to the target.