We now have a pair of equations for each point in the image. It would be very costly to solve these equations simultaneously by one of the standard methods, such as Gauss-Jordan elimination [ll, 131. The corresponding matrix is sparse and very large since the number of rows and columns equals twice the number of picture cells in the image. Iterative methods, such as the GaussSeidel method [ll, 131, suggest themselves. We can compute a new set of velocity estimates (un+', vn+') from the estimated derivatives and the average of the previous velocity estimates (un, vn) by