Advertisement

Occluded objects recognition using multiscale features and Hopfield neural networks

  • Jiann-Shu Lee
  • Chin-Hsing Chen
  • Yung-Nien Sun
  • Guan-Shu Tseng
Neural Networks
Part of the Lecture Notes in Computer Science book series (LNCS, volume 974)

Abstract

A new method to recognize partially visible two-dimensional objects by means of multiscale features and Hopfield neural network is proposed. The Hopfield network is employed to perform global feature matching. Since the network only guarantee to converge to a local optimal state, the matching results heavily depend on the initial network state determined by the extracted features. To acquire more satisfactory initial matching results, a new feature vector, consisting of the multiscale evolution of the extremal position and magnitude of the wavelet transformed contour orientation, is developed. These features contain the contour primitives information in a multiscale manner, hence good initial states can be obtained. The good initiation enables the method to recognize objects of even heavily occluded, that can not be achieved by using the Nasrabadi-Li's method. In addition, to make the matching results more insensitive to the threshold value selection of the network, we replace the step-like thresholding function by a ramp-like one. Experimental results have shown that our method is effective even for noisy occluded objects.

References

  1. 1.
    J. L. Turney, T. N. Mudge and R. A. Voltz, “Recognizing partially occluded parts,” IEEE Trans. Pattern Analysis Mach. Intell. PAMI-7, pp. 410–421, 1985.Google Scholar
  2. 2.
    H. C. Liu and M. D. Srinath, “Partial shape classification using contour matching in distance transformation,” IEEE Trans. Pattern Analysis Mach. Intell. PAMI-12, pp. 1072–1079, 1990.Google Scholar
  3. 3.
    N. Ansari and E. J. Delp, “Partial shape recognition: a landmark-based approach,” IEEE Trans. Pattern Anal. Mach. Intell. PAMI-12, pp. 470–483, 1990.Google Scholar
  4. 4.
    Nasser M. Nasrabadi, and Wei Li, “Object recognition by a Hopfield neural network,” IEEE Trans. on Syst. Man Cybern. SMC-21, pp. 1523–1535, 1991.Google Scholar
  5. 5.
    J. S. Lee, Y. N. Sun, C. H. Chen, and C. T. Tsai, “Wavelet based corner detection,” Pattern Recognition, Vol. 26, No. 6, pp. 853–865, 1993.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1995

Authors and Affiliations

  • Jiann-Shu Lee
    • 1
  • Chin-Hsing Chen
    • 1
  • Yung-Nien Sun
    • 2
  • Guan-Shu Tseng
    • 1
  1. 1.Department of Electrical EngineeringNational Cheng Kung UniversityTainanTaiwan, R.O.C.
  2. 2.Institute of Information EngineeringNational Cheng Kung UniversityTainanTaiwan, R.O.C.

Personalised recommendations