A robust tracking of 3D motion
The tracking of moving objects in the 3D space for long-term image sequences must be very robust with respect to noise and computational errors. Thus, for example autoregressive, and Newtonian models have been adopted mainly with least-square, Kalman filter, and other techniques. The parameters measured are predicted/corrected on the basis of the model adopted; which can be adaptive or not. In this paper, a new method for tracking objects in the 3D space belonging to the class of matching-based algorithms with an adaptive prediction/correction mechanism is presented. The prediction/correction is based on 2D and 3D motion estimations, and both these corrections are used for measuring the displacements on the image plane. The mechanism proposed is very robust with respect to the accumulation error and, thus, it is suitable for very long-term object tracking.
Unable to display preview. Download preview PDF.
- 1.C. Longuet-Higgins and K. Prazdny, “The interpretation of a moving retinal image,” Proc. Roy. Soc. London B, vol. 208, pp. 385–397, 1980.Google Scholar
- 2.B. K. P. Horn and B. G. Schunck, “Determining optical flow,” Artificial Intelligence, vol. 17, pp. 185–203, 1981.Google Scholar
- 3.P. Nesi, “Variational approach for optical flow estimation managing discontinuities,” Image and Vision Computing, vol. 11, no. 7, pp. 419–439, 1993.Google Scholar
- 4.D. Ben-Tzvi, A. DelBimbo, and P. Nesi, “Optical flow from constraint lines parametrization,” Pattern Recognition, vol. 26, pp. 1549–1561, 1993.Google Scholar
- 5.H. G. Musmann, P. Pirsh, and H.-J. Grallert, “Advanced in picture coding,” Proceedings of the IEEE, vol. 73, pp. 523–548, 1985.Google Scholar
- 6.H. Shariat and K. E. Price, “Motion estimation with more than two frames,” IEEE Trans. on Pat. Ana. and Mac. Intel., vol. 12, pp. 417–434, 1990.Google Scholar
- 7.H. Li, P. Roivainen, and R. Forchheimer, “3-d motion estimation in model-based facial image coding,” IEEE Trans. on Pat. Ana. and Mac. Intel., vol. 15, pp. 545–555, 1993.Google Scholar
- 8.T. J. Broida and R. Chellappa, “Experiments and uniqueness results on object structure and kinematics from a sequence of monocular images,” in Proc. of the IEEE Workshop on Visual Motion, Irvine, USA, pp. 21–30, 1989.Google Scholar
- 9.A. DelBimbo and P. Nesi, “Behavioral object recognition from multiple image frames,” Signal Processing, vol. 27, no. 1, pp. 37–49, 1992.Google Scholar
- 10.D. Terzopoulos and K. Waters, “Analysis and synthesis of facial image sequences using physical and anatomical models,” IEEE Trans. on Pat. Ana. and Mac. Intel., vol. 15, pp. 569–579, 1993.Google Scholar
- 11.R. Forchheimer and T. Kronander, “Image coding — from waveforms to animation,” IEEE Trans. on Acou., Speech, and Sig. Proc., vol. 37, pp. 2008–2023, 1989.Google Scholar
- 12.R. Koch, “Dynamic 3-d scene analysis through synthesis feedback control,” IEEE Trans. on Pat. Ana. and Mac. Intel., vol. 15, pp. 556–568, 1993.Google Scholar
- 13.M. Rydfalk, “Candide, a parameterised face,” tech. rep., Department of Electrical Engineering, Linköping University, LiTH-ISY-I-0866, Sweden, Oct 1987.Google Scholar