not sure who was it that said that 3d tracking doesn't work with cameras. at university i have worked with a toolkit that does 3d position tracking based on webcam pictures, you just require fixed image tags that can be detected in the picture and then you can calculate the angle under which the tag picture is seen as well as the position of it.
(here is a link to one of the toolkits working with this: http://studierstube.icg.tu-graz.ac.at/handheld_ar/stbtracker.php )
but when lookig more on the patent text it seems that they are going to track just the 2 dimensional position of the spheres (the patent also mentions that it is not limited to perfect speres) in the image which coresponds to a line in 3d space and se ultrasonic to detect the distance of the controller (so you can detect on which position of the line the controller is). this would allow full 3d tracking of the controller position but the resolution will still depend on the resolution of the visual tracking.
to me it looks like a possible way to implement full 3d tracking with rather low processing power requirements.
finally i want to say that sony patenting this doesn't mean it will ever get released to the public in this form, and if it gets released its still not decided if they will use it for ps3 or if it will release with ps4.