ESTIMATION OF OPTICAL-FLOW BASED ON HIGHER-ORDER SPATIOTEMPORAL DERIVATIVES IN INTERLACED AND NONINTERLACED IMAGE SEQUENCES

This contribution(1) investigates local differential techniques for es timating optical flow and its derivatives based on the brightness chan ge constraint. By using the tensor calculus representation we build th e Taylor expansion of the gray-value derivatives as well as of the opt ical flow in a spatiotemporal neighborhood. Such a formulation provide s a unifying framework for all existing local differential approaches and allows to derive new systems of equations for the estimation of th e optical flow and of its derivatives. We also tested various optical flow estimation approaches on real image sequences recorded by a calib rated camera which was fixed on the arm of a robot. By moving the arm of the robot along a precisely defined trajectory, we can determine th e true displacement rate of scene surface elements projected into the image plane and compare it quantitatively with the results of differen t optical flow estimators. Since the optical flow estimators are based on gray-value derivatives of up to fourth-order, we were forced to de velop modified Gaussian derivative filters to obtain acceptable estima tes for the derivatives, Further, we show quantitatively that these fi lters contribute to a much more robust optical how estimation, In addi tion, successive lines of TV-cameras have an offset in time due to the interlace technique. We demonstrate the adaptation of filter kernels for estimating higher-order spatiotemporal derivatives in interlaced i mage sequences.