A MODEL FOR VISUAL IMAGE-BACKGROUND DISCRIMINATION BY RELATIVE MOVEMENT

Biologically plausible electronic neural network setup for real time p rocessing motion image information was built. Using this setup the fir st part of the model was examined and real rime discrimination of movi ng object image was realized from complex background in high resolutio n. Afterimages may play an important role in filtering moving object i mage and the aperture problem should be separated into two parts: the first part, i.e. the incomplete filtered moving object image, can be b etter resolved by parallel integration of multi-channel visual informa tion, however, the second part, i.e. the inaccurate measurement result s for movement direction, may only get certain compensation by visual integration.