Spatio-temporal Fourier transforms for visual discrimination

Intelligent pattern recognition techniques have become important in the fields of computer and robot vision. Our interest focuses on detection of dynamical variables which allows robotic visual units to define object evolution from 2D images. We describe a computational algorithm which predicts the real scene images using spatial Fourier frequency transforms multiplied by transforms of dynamic model integro-differential equations. The theoretical results have been simulated by computer numerical image processing and compared to expected real scenes.