Abstract
The relaxation labeling is a useful technique to deal with local ambiguity and achieve consistency. In [1.], some useful comments indicate several common properties exist in the relaxation process and the neural network technique. Neural networks can be used as an efficient tool to optimize the average local consistency function whose optimal solution results in a compatible label assignment. However, most of current investigations in this field are based on the standard Hopfield neural network (SHNN) presented in [2.]. In this paper, an improved Hopfield neural network (IHNN) presented in [3.] is utilized to fulfill relaxation labeling. Compared to the SHNN, this approach has some advantages. 1) The IHNN uses fewer neurons than that of SHNN. 2) The activation function of IHNN is easier to be implemented than that of SHNN. 3) The IHNN does not contain any penalty parameters. It can generate the exact optimal solution. Some experimental results illustrate that the IHNN approach can obtain a better labeling performance than that of SHNN.
Access provided by Autonomous University of Puebla. Download to read the full chapter text
Chapter PDF
Similar content being viewed by others
Keywords
- Neural Network
- Penalty Parameter
- Chinese Character
- Neural Network Approach
- Quadratic Optimization Problem
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
References
Yu, S. S., Tsai, W. H.: Relaxation by the Hopefield Neural Network. Pattern Recongition. 25 (1992) 197–209
Tank, D. W., Hopfield, J. J.: Simple Neural Optimization Networks: an A/D Converter, Signal Decision Circuit and a Linear Programming Circuit. IEEE Transcations on Circuits and Systems. 35 (1988) 554–562
Liang, X. B., Wang, J.: A Recurrent Neural Network for Nonlinear Optimization with a Continuously Differentiable Objective Function and Bound Constraints. IEEE Transcations on Neural Networks. 11 (2000) 1251–1262
Rosenfeld, A., Kak, A. C.: Digital Picture Processing Volume 2. New York: Academic Press, (1982)
Rosenfeld, A., Hummel, R. A., Zucker, S. W.: Scene Labeling by Relaxation Operations. IEEE Transcations on System, Man and Cybernetics. 6 (1976) 420–433
Peleg, S., Rosenfeld, A.: Determining Compatibility Coefficients for Curve Enchancement Relaxation Processes. IEEE Transcations on System, Man and Cybernetics. 8 (1978) 548–555
Hummel, R. A., Zucker, S. W.: On the Foundations of Relaxation Labeling. IEEE Transcations on Pattern Analysis and Machine Intelligence. 5 (1983) 267–287
Kennedy, M. P., Chua, L. O.: Neural Networks for Nonlinear Programming. IEEE Transcations on Circuits and Systems Part A. 35 (1988) 554–562
Zhang, Y. N., Wang, J., Xia, Y. S.: A Dual Neural Network for Redundancy Resolution of Kinematically Redundant Manipulators Subject to Joint Limits and Joint Velocity Limits. IEEE Transcations on Neural Networks. 14 (2003) 658–667
Xia, Y. S., Wang, J.: A Recurrent Reural Network for Solving Nonlinear Convex Programs Subject to Linear Constraints. IEEE Transcations on Neural Networks. 16 (2005) 379–386
Sang, N., Zhang, T. X.: Segmentation of FLIR Images by Hopfield Neural Network with Edge Constraint. Pattern Recongition. 34 (2001) 811–821
Kurugollu, F., Sankur, B., Harmanci, A. E.: Image Segmentation by Relaxation Using Constraint Satisfaction Neural Network. Image and Vision Computing. 20 (2002) 483–497
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2006 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Cheng, L., Hou, ZG., Tan, M. (2006). Relaxation Labeling Using an Improved Hopfield Neural Network. In: Huang, DS., Li, K., Irwin, G.W. (eds) Intelligent Computing in Signal Processing and Pattern Recognition. Lecture Notes in Control and Information Sciences, vol 345. Springer, Berlin, Heidelberg . https://doi.org/10.1007/978-3-540-37258-5_44
Download citation
DOI: https://doi.org/10.1007/978-3-540-37258-5_44
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-37257-8
Online ISBN: 978-3-540-37258-5
eBook Packages: EngineeringEngineering (R0)