Abstract

As a person moves through the environment, he or she must be able to judge heading and detect moving objects in order to intercept or avoid them.  For a moving observer, the retinal images of stationary objects move in a pattern known as the optic flow field.  Observers are able to use this optic flow field to judge heading and detect moving objects accurately under a variety of conditions.  I will present a computational model that makes use of operators that are tuned for direction and speed of motion, similar to the responses of neurons in area MT of the visual cortex, to compute observer heading and detect the presence of moving objects.  I will also show that adding tuned stereo responses can enhance the model's ability to distinguish stationary from moving objects in the scene.