Hierarchical pattern recognition method with the multi-level distance function
To overcome the disadvantages of the conventional distance pattern recognition method, based on the distance function of a complicated form, it is proposed to introduce a set of simpler functions which form a multi-level distance function. The resulting recognition method consists in gradual elimination of the classes to which a recognized object does not belong. The proposed, very simple algorithm operates in such a way as if in the set of classes there existed a hierarchy which would make it possible to exclude subsequently the whole groups of classes; however, such hierarchy is not introduced. It is (implicitly) comprised in the structure of the recognition algorithm. This provides for a significant reduction of the execution time. The algorithm has been applied to object recognition in computer vision, where the requirement of large calculation speed is particularly important.
References
H.C. Barrow, J. Tennenbaum, Computational vision, Proc. IEEE, 69, 5, 572-593, 1981.
E. Charniak, D. Mcdermott, Introduction to artificial intelligence, Addison-Wesley Publishing Company 1985.
W.B. Gevarter, Intelligent machines: an introductory perspective of artificial intelligence and robotics, Englewood Cliffs, Prentice-Hall 1985.
M. Brady, Computer vision, Artificial Intelligence, 19, 1, 7-16, 1982.
N.R. Corsv Jr, Machine vision for robotics, IEEE Trans., 1E-30, 3, 282-291, 1983.
D. Marr, Vision, Freeman, San Francisco 1982.
D. Ballard, C. Brown, Computer vision, Englewood Cliffs, Prentice-Hall 1982.
A. Rosenfeld, A. Kak, Digital picture processing, vol. 1 and 2, Academic Press, New York 1982.
J. Malec, M. Wnuk, Artificial intelligence in robotics, in: Fundamental Problems of Technology, vol.25, A.Morecki [Ed.]: Robotics [in Polish], PWN.
L.J. Chmielewski, Robotics, vision systems, expert systems and artificial intelligence, in: Computer Modelling of Perception and Reasoning, W. Kosiński [Ed.], [in Polish], IFTR Reports 31/1987, pp.6-72.
J.T. Tou, R.C. Gonzalez, Pattern recognition principles, Advanced Books Program, Addison-Wesley Publishing Comp., Inc., Reading, Massachusetts 1974.
L.J. Chmielewski, Time-optimal recognition method with the multi-level distance function, Proc. 3rd Int. Conf. on Machine Vision and its Applications, Warwick, UK, July 21-23, 1989; IEE Publication No. 307.
L.J. Chmielewski, Feature selection and time minimization in the multi-level minimum - distance recognition algorithm, Proc. 9th Conf. on Computer Methods in Mechanics, Rytro 16-20 May 1989 (Late Papers), Wyd. Politechniki Krakowskiej, 1989.
L.J. Chmielewski, Proposition of the standard way of naming variables in the image analyzing software, in: Computer Modelling of Perception and Reasoning - II, L.J. Chmielewski, W.Kosiński [Eds.], [in Polish], IFTR Report 42/1988, pp. 80-92.
L.J. Chmielewski, W. Kosiński, Software for recognizing two-dimensional, fully visible objects in binary pictures [in Polish], Scientific Report SCSS, TDC IFTR PAS, 1987.