H. Taichi Wang

Computing optical flow in the primate visual system

Computing motion on the basis of the time-varying image intensity is a difficult problem for both artificial and biological vision systems. We show how gradient models, a well-known class of motion algorithms, can be implemented within the magnocellular pathway of the primates visual system. Our cooperative algorithm computes optical flow in two steps. In the first stage, assumed to be located in primary visual cortex, local motion is measured while spatial integration occurs in the second stage, assumed to be located in the middle temporal area (MT). The final optical flow is extracted in this second stage using population coding, such that the velocity is represented by the vector sum of neurons coding for motion in different directions. Our theory, relating the single-cell to the perceptual level, accounts for a number of psychophysical and electrophysiological observations and illusions.

Analog neural networks for focal-plane image processing

Autonomous systems have to overcome two major problems in interpreting the data collected by sensors. These problems are due to the fact that multiple real world scenes can produce a given intensity distribution and that sensor data is often corrupted by noise. These problems can be solved by imposing certain a priori knowledge about the world to arrive at a unique solution. This reduces early vision problems to constrained optimization problems, which can be conveniently formulated as energy minimization problems in the framework of Regularization Theory. These energy functions can be implemented in analog hardware. This feature makes this approach very attractive for real time autonomous systems. We are presently developing algorithms and analog CMOS chips based on this approach for focal plane image processing.