It seems that people pay particular attention to certain movements (it’s recently been suggested that Autists develop a preference for coordinated movement and sound, for example).
Simple linear movement is easy to predict and so our eyes automatically jump to their best estimate of where it will end up. More complex motion requires us to pay a great deal more attention to it.
When there are both simple and complex motions in a scene, eg when driving, we will therefore tend to completely miss certain curvilinear movements -this is probably a cause of those accidents in which people say “I didn’t see him until it was too late…He came out of nowhere”.
Today’s invention is an image processing algorithm which undertakes a realtime analysis on the restricted visual field into which a vehicle is about to move (with emphasis on the margins -from which objects will tend to emerge). It applies directional (Sobel-esque) filters and, when a new curvilinear motion is detected, it issues a “psst”-type warning to the driver.