Moving object nonlinear trajectory estimation in the transform/spatiotemporal mixed domain

Presents a discrete transform/spatiotemporal mixed domain nonlinear trajectory estimation algorithm. The trajectory of an object moving in an image sequence is modeled by quadratic polynomials. The application of the discrete Fourier transform along the two spatial dimensions of the image sequence results in one dimensional chirp sequences along the temporal dimension. The parameters of the chirp sequences are estimated using a well known chirp signal parameter estimation method. From the estimates of the chirp sequence parameters, the coefficients of the quadratic polynomials and, hence, the nonlinear trajectory coefficients are determined. Examples of moving object quadratic trajectory estimation are given