Each star will give a circle on which you could be located. 2 will be the intersection of 2 circles, which is 2 points. If you have some additional assumptions (e.g. Northern hemisphere), it is possible to eliminate one of the points. Otherwise you need a 3rd point to get a full triangulation.
What you're describing is a lat/lon position fix, but the link nickponline gave is for a plate solver (finding the RA/Dec of stars in an image, and determining the camera orientation). All other plate solvers I've looked at use three or four stars and measure the angles and distances relative to each other. Since nickponline's code appears to only use two, I'm intrigued as to what the angle is being measured relative to. It kinda looks like it requires the image to be oriented with North as up, which seems like an interesting twist. It seems a little like cheating, but honestly most people will know the orientation of their camera, and it probably speeds up the search considerably, and would reduce/eliminate the need for building index files.
