This Article 
 Bibliographic References 
 Add to: 
A Minimum Description Length Model for Recognizing Objects with Variable Appearances (The VAPOR model)
October 1994 (vol. 16 no. 10)
pp. 1032-1036

Most object recognition systems can only model objects composed of rigid pieces whose appearance depends only on lighting and viewpoint. Many real world objects, however, have variable appearances because they are flexible and/or have a variable number of parts. These objects cannot be easily modeled using current techniques. The author proposes the use of a knowledge representation method called the VAPOR (Variable APpearance Object Representation) model to represent objects with these kinds of variable appearances. The VAPOR model is an idealization of the object; all instances of the model in an image are variations from the ideal appearance. The variations are evaluated by the description length of the data given the model, i.e., the number of information-theoretic bits needed to represent the model and the deviations of the data from the ideal appearance. The shortest length model is chosen as the best description. The author demonstrates how the VAPOR model performs in a simple domain of circles and polygons and in the complex domain of finding cloverleaf interchanges in aerial images of roads.

[1] M. Boddy and T. Dean, "Solving time-dependent planning problems," inProc. Int. Joint Conf. Artificial Intell., 1989, pp. 979-984.
[2] R. A. Brooks, Symbolic reasoning among 3-D models and 2-D images,"Artificial Intell., vol. 17, pp. 285-348, 1981.
[3] J. Canning, "Recognizing objects with Variable APpearance--The VAPOR system," Tech. Rep. CAR-TR-563, CS-TR-2705, Center for Automat. Res., July 1991.
[4] P. Fua and A. J. Hanson, "Objective functions for feature discrimination," inProc. Int. Joint Conf. Artificial Intell., 1989, pp. 1596-1603.
[5] W. E. L. Grimson, "Recognition of object families using parameterized models," inProc. Int. Conf. Comput. Vision, 1987, pp. 93-101.
[6] G. Kanizsa, "Subjective contours,"Scientific Amer., vol. 234, pp. 48-52, 1976.
[7] M. Kass, A. Witkin, and D. Terzopoulos, "Snakes: Active contour models,"Int. J. Comput. Vision, vol. 1, pp. 321-331, 1988.
[8] D. J. Kriegman and J. Ponce, "On recognizing and positioning 3- D objects from image contours,"IEEE Trans. Pattern Anal. Machine Intell., vol. 12, pp. 1127-1137, Dec. 1990.
[9] D. G. Lowe, "The viewpoint consistency constraint,"Int. J. Comput. Vision, vol. 1, pp. 57-72, 1987.
[10] E. P. D. Pednault, "Some experiments in applying inductive inference principles to surface reconstruction," inProc. Int. Joint Conf. Artificial Intell., 1989, pp. 1603-1609.
[11] J. Rissanen, "Modeling by shortest data description,"Automatica, vol. 14, pp. 465-471, 1978.
[12] J. Rissanen, "A universal prior for integers and estimation by minimum description length,"Ann. Statist., vol. 11, pp. 416-431, 1983.
[13] J. Segen, "Model learning and recognition of nonrigid objects," inProc. CVPR '89, San Diego, CA, pp. 597-602.
[14] D. Terzopoulos, A. Witkin, and M. Kass, "Symmetry-seeking models and 3D object reconstruction,"Int. J. Comput. Vision, vol. 1, pp. 211-221, 1987.

Index Terms:
image recognition; knowledge representation; minimum description length model; VAPOR model; object recognition systems; knowledge representation method; variable appearance object representation model; ideal appearance; shortest length model; cloverleaf interchanges; aerial images; roads
J. Canning, "A Minimum Description Length Model for Recognizing Objects with Variable Appearances (The VAPOR model)," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 10, pp. 1032-1036, Oct. 1994, doi:10.1109/34.329006
Usage of this product signifies your acceptance of the Terms of Use.