Sensitivity enhancement of elementary velocity estimators with self and lateral facilitation mechanisms

Elementary velocity estimators (EVEs), i.e. neural network units designed to locally extract the direction and speed of motion from visual information, exhibit limitations in the minimum time-separation between successive motion events. This is due to inhibitory activity required for proper operation of an EVE over a specific speed range. These limitations may be relaxed by incorporating self and lateral facilitation mechanisms. These mechanisms are incorporated into the elementary motion detector portion of an EVE. Two different configurations are proposed. One configuration involves only lateral facilitation schemes, while the other involves both self and lateral schemes. Self and lateral facilitation schemes are shown to enhance the sensitivity of EVEs to successive motion events with small time separation, while the magnitude and the extent of the inhibitory activity required for a specific speed range is maintained.<>