Assuming this statement is true, then why is it often difficult to make out suspects in security camera footage shown on TV news reports? The classic example is a case involving a theft or other crime at a mall -- facial features are never clear. Banks, however, seem to always have good images available when robberies take place. Is it an issue of placement, older technologies, or something else?
The vast majority of CCTV cameras sold today are D1 (VGA) resolution. They are also often covering fairly large fields of view, so you end up with a few dozen pixels on target of someones face, or a license plate. It's impossible (other than on TV) to extrapolate that into useful detail.
The "motion detection" systems described in this article are also false-alarm generating garbage in most cases. You need a camera that can definitively recognize and detect people, plus have some ability to a behavior analysis (direction of travel, loitering,etc.).
An example from an actual camera on my driveway (with the Pixim CMOS sensor): http://www.youtube.com/videoiq#p/u/6/FbAWYB24Qdw
Track the person. Not the dogs, not the water, not the shadows.
This is a very tricky problem and there is a lot of work being done in this area. Surprisingly one of the biggest issues is getting customers to realize that there is a real benefit to upgrading their cameras.