A new mathematical human vision model with an autonomous image observing mechanism and its application to multiple motion detection

A spatio-temporal human vision model which can simulate various kinds of perceptual responses in human vision has already been proposed. As a first step to develop more advanced pattern recognition systems or computer vision systems, this paper focuses on formulating a new human vision model with a multiple motion detection ability by means of appropriately improving the former model. As a pertinent example, the new model is applied to the analysis of a compound wave consisting of three different drifting sinusoidal waves. It is shown that the new model is very effective both in reproducing various kinds of perceptual responses to motion images, and in estimating each component wave's motion velocity.