목차
1. Introduction
2. Problem Formulation
3. Determining Energy Function
4. Minimizing the Energy using a Neural Network
5. Application to Boundary Detection
7. Conclusion
2. Problem Formulation
3. Determining Energy Function
4. Minimizing the Energy using a Neural Network
5. Application to Boundary Detection
7. Conclusion
본문내용
oduces better boundaries; the boundaries are smooth and correct.
Fig. 8 (a) and (b) upper and lower boundaries taken by plain Kalman filter.
Fig. 9 (a) and (b) upper and lower boundaries of the test images obtained by the proposed method.
7. Conclusion
This paper introduced a boundary tracking algorithm that consists of data association and a tracking filter. We applied the basic concept of the data association technique proposed in the field of multiple target tracking, to detecting object boundaries. After defining the data association as a constrained optimization problem, we introduced a new energy function and a Hopfield network as an efficient method for solving the energy function. The new algorithm shows good performance in finding object boundaries especially in IR images. An open problem is to include the constraints of parallel boundary condition for the upper and lower boundaries. Also another problem is to determine the system parameter included in the energy function automatically.
References
(1) D. L. Alspach, A Gaussian sum approach to multi-target identification tracking problem, Automatica, 11, pp. 285 - 296, 1975.
(2) Y. Bar-Shalom, Extension of probabilistic data association filter in multi-target tracking, Proc. 5th Symp. Nonlinear Estimation Theory and its Application, pp. 16 - 21, 1974.
(3) T. E. Fortmann and Y. Bar-Shalom, Tracking and Data Association, Orlando Academic Press, Inc, 1988.
(4) T. E. Fortmann, Y. Bar-Shalom and M. Scheffe, Sonar tracking of multiple targets using joint probability data association, IEEE Journal of Oceanic Engineering, 8, pp. 173 - 183, 1983.
(5) Kuczewski, R., Neural network approaches to multi-target tracking, Proc. of the IEEE, ICNN Conference.
(6) D. B. Reid, An algorithm for tracking multiple targets, IEEE Trans. on Automatic Control, 24, pp. 843 - 854, 1979.
(7) D. Sengupta and R. A. Iltis, Neural solution to the multi-target tracking data association problem, IEEE Trans. on Aerospace and Electronic Systems, 25, pp. 96 - 108, 1989.
(8) P. Smith and G. Buechler, A branching algorithm for discrimining and tracking multiple objects, IEEE Trans. on Automatic Control, 20, pp. 101 - 104, 1975.
(9) H. M. Sun and S. M. Chiang, Tracking multi-target in cluttered environment, IEEE Trans. on Aerospace and Electronic System, 28, pp. 546 - 559, 1992.
(10) Y. W. Lee and H. Jeong, A Neural Network Approach tp the Optimal Data Association in Multi-Target Tracking, Proc. of WCNN'95, INNS Press, 1995.
(11) J. G. Kim and H. Jeong, Neural Networks for Optimal Data Association of Hot Plate, Proc. of WCNN'95, INNS Press, 1995.
(12) B. Park, M. Chun, J. Kim, J. Yi, H. Chung, and S. Hwang, On-line control of front end bending in plate rolling, 2nd European Conference on Continuous Casting, 6th International Rolling Conference on Flat Products, Dsseldorf, June, 1994.
(13) J. J. Hopfield and D. W. Tank, Neural computation of decisions in optimization problems, Biological Cybernetics, 52, pp. 141 - 152, 1985.
Fig. 8 (a) and (b) upper and lower boundaries taken by plain Kalman filter.
Fig. 9 (a) and (b) upper and lower boundaries of the test images obtained by the proposed method.
7. Conclusion
This paper introduced a boundary tracking algorithm that consists of data association and a tracking filter. We applied the basic concept of the data association technique proposed in the field of multiple target tracking, to detecting object boundaries. After defining the data association as a constrained optimization problem, we introduced a new energy function and a Hopfield network as an efficient method for solving the energy function. The new algorithm shows good performance in finding object boundaries especially in IR images. An open problem is to include the constraints of parallel boundary condition for the upper and lower boundaries. Also another problem is to determine the system parameter included in the energy function automatically.
References
(1) D. L. Alspach, A Gaussian sum approach to multi-target identification tracking problem, Automatica, 11, pp. 285 - 296, 1975.
(2) Y. Bar-Shalom, Extension of probabilistic data association filter in multi-target tracking, Proc. 5th Symp. Nonlinear Estimation Theory and its Application, pp. 16 - 21, 1974.
(3) T. E. Fortmann and Y. Bar-Shalom, Tracking and Data Association, Orlando Academic Press, Inc, 1988.
(4) T. E. Fortmann, Y. Bar-Shalom and M. Scheffe, Sonar tracking of multiple targets using joint probability data association, IEEE Journal of Oceanic Engineering, 8, pp. 173 - 183, 1983.
(5) Kuczewski, R., Neural network approaches to multi-target tracking, Proc. of the IEEE, ICNN Conference.
(6) D. B. Reid, An algorithm for tracking multiple targets, IEEE Trans. on Automatic Control, 24, pp. 843 - 854, 1979.
(7) D. Sengupta and R. A. Iltis, Neural solution to the multi-target tracking data association problem, IEEE Trans. on Aerospace and Electronic Systems, 25, pp. 96 - 108, 1989.
(8) P. Smith and G. Buechler, A branching algorithm for discrimining and tracking multiple objects, IEEE Trans. on Automatic Control, 20, pp. 101 - 104, 1975.
(9) H. M. Sun and S. M. Chiang, Tracking multi-target in cluttered environment, IEEE Trans. on Aerospace and Electronic System, 28, pp. 546 - 559, 1992.
(10) Y. W. Lee and H. Jeong, A Neural Network Approach tp the Optimal Data Association in Multi-Target Tracking, Proc. of WCNN'95, INNS Press, 1995.
(11) J. G. Kim and H. Jeong, Neural Networks for Optimal Data Association of Hot Plate, Proc. of WCNN'95, INNS Press, 1995.
(12) B. Park, M. Chun, J. Kim, J. Yi, H. Chung, and S. Hwang, On-line control of front end bending in plate rolling, 2nd European Conference on Continuous Casting, 6th International Rolling Conference on Flat Products, Dsseldorf, June, 1994.
(13) J. J. Hopfield and D. W. Tank, Neural computation of decisions in optimization problems, Biological Cybernetics, 52, pp. 141 - 152, 1985.