How does Google do Maps' Street View Cursor "Following"

Question

In Google Maps Street View your cursor turns into a rectangular/oval shape as you mouse over different parts of the scene. For example:

http://maps.google.com/?q=loc:+Maryland+Ave+at+e.+26th+st+Baltimore+MD+US&ie=UTF8&z=16&iwloc=A&layer=c&cbll=39.319313,-76.618426&panoid=6W2XgkHoGuf6_SKv0LIL9Q&cbp=12,307.06,,0,3.16

As you move the cursor over the building it "hugs" the walls. It's not just as simple as following the intersection because if you continue on to the left you can see the angle change as it hits different faces of the buildings.

Do they do some sort of image analysis to identify faces of the buildings or do they, as they take the picture, do some sort of laser range finder and then later combine it with the picture?

Answer 1

They do use laser range scanners. And according to the Google Lat Long Blog:

We have been able to accomplish this by making a compact representation of the building facade and road geometry for all the Street View panoramas using laser point clouds and differences between consecutive pictures.