This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
Learning and Feature Selection in Stereo Matching
September 1994 (vol. 16 no. 9)
pp. 869-881

We present a novel stereo matching algorithm which integrates learning, feature selection, and surface reconstruction. First, a new instance based learning (IBL) algorithm is used to generate an approximation to the optimal feature set for matching. In addition, the importance of two separate kinds of knowledge, image dependent knowledge and image independent knowledge, is discussed. Second, we develop an adaptive method for refining the feature set. This adaptive method analyzes the feature error to locate areas of the image that would lead to false matches. Then these areas are used to guide the search through feature space towards maximizing the class separation distance between the correct match and the false matches. Third, we introduce a self-diagnostic method for determining when apriori knowledge is necessary for finding the correct match. If the a priori knowledge is necessary then we use a surface reconstruction model to discriminate between match possibilities. Our algorithm is comprehensively tested against fixed feature set algorithms and against a traditional pyramid algorithm. Finally, we present and discuss extensive empirical results of our algorithm based on a large set of real images.

[1] D. W. Aha and D. Kibler, "Noise-tolerant instance-based learning algorithms," inProc. Int. Joint Conf. Artificial Intell., 1989, pp. 794-799.
[2] S. T. Barnard, "Stereo matching by hierarchical, microcanonical annealing," inIEEE Proc. Image Understanding Workshop, Feb. 1987.
[3] L. Cohen, L. Vinet, P. Sander, and A. Gagalowicz, "Hierarchical region based stereo matching," inProc. Comput. Vision Pattern Recognit., 1989, pp. 416-421.
[4] J. D. Cowan and D. H. Sharp, "Neural nets and artificial intelligence,"Daedalus, vol. 117, no. 1, pp. 85-121, 1988.
[5] P. A. Devijver and J. Kittler,Pattern Recognition: A Statistical Approach. Englewood Cliffs, NJ: Prentice Hall, 1982.
[6] W. E. L. Grimson,From Images to Surfaces: A Computational Study of the Human Early visual System. Cambridge, MA: MIT Press, 1981.
[7] E. Gulch, "Results of test on image matching of ISPRS WG III/4,"Int. Archives of Photogrammetry and Remote Sensing, vol. 27-111, pp. 254-271, 1988.
[8] J. G. Harris, "A new approach to surface reconstruction: The coupled depth/slope model," inFirst Int. Conf. Comput. Vision, 1987, pp. 277-283.
[9] M. J. Hannah, "A system for digital stereo image matching,"Photogrammetric Eng. Remote Sensing, vol. 55, no. 12, pp. 1765-1770, Dec. 1989.
[10] R. M. Haralick and L. G. Shapiro,Computer and Robot VisionReading, MA: Addison-Wesley, vol. II, 1993, pp. 357-361.
[11] W. A. Hoff and N. Ahuja, "Depth from stereo," inProc. Fourth Scandinavian Conf. Image Anal., Trondheim, Norway, June 18-20, 1985, pp. 761-768.
[12] J. Kittler, "Une generalisation de quelques algorithmes sous-optimaux de recherche d'ensembles d'attributs," inProc. Reconn. des Formes et Trait. des Images, Paris, 1978, pp. 678-686.
[13] H. S. Lim and T. O. Binford, "Stereo correspondence: A hierarchical approach," inIEEE Proc. Image Understanding Workshop, Feb. 1987, pp. 234-241.
[14] S. B. Marapane and M. M. Trivedi, "Multi-primitive hierarchical (MPH) stereo system,"IEEE Proc. Comput. Vision Pattern Recognit., 1992, pp. 499-505.
[15] T. Marill and D. M. Green, "On the effectiveness of receptors in recognition systems,"IEEE Trans. Inform. Theory, vol. IT-9, pp. 11-17, Jan. 1963.
[16] D. Marr and T. Poggio, "A theory of edge detection," inJ. Roy. Soc. Lon. B, vol. 204, pp. 301-328, 1980.
[17] D. M. McKeown and Y. C. Hsieh, "Hierarchical waveform matching: A new feature-based stereo technique," inIEEE Proc. Comput. Vision and Pattern Recognit., 1992, pp. 513-519.
[18] H. P. Moravec, "Obstacle avoidance and navigation in the real world by a seeing robot rover," Ph.D. dissertation, Stanford Univ, Stanford, CA, Sept. 1980.
[19] P. M. Narendra and K. Fukunaga, "A branch and bound algorithm for feature subset selection,"IEEE Trans. Comput., vol. C-26, pp. 917-922, Sept. 1977.
[20] S. Negahdaripour and A. K. Jain, "Challenges in computer vision: future research directions,"IEEE Proc. Comput. Vision Pattern Recognition, pp. 189-199, 1992.
[21] D. Terzopoulos, "Multi-level computational processes for visual surface reconstruction,"Comput. Vision, Graphics, Image Processingvol. 24, pp. 52-96, 1983.
[22] D. Terzopoulos, "The computation of visible-surface reconstruction,"IEEE Trans. Pattern Anal. Machine Intell., vol. 10, no. 4, pp. 417-437, 1988.
[23] J. N. Weng, N. Ahuja, and T. S. Huang, "Matching two perspective views,"IEEE Trans. Pattern Anal. Machine Intell., vol. 14, no. 8, pp. 806-825, 1992.
[24] A. Whitney, "A direct method of nonparametric measurement selection,"IEEE Trans. Comput., vol. C-20, pp. 1100-1103, 1971.

Index Terms:
feature extraction; stereo image processing; error analysis; learning (artificial intelligence); adaptive systems; image reconstruction; feature selection; stereo matching; instance based learning; surface reconstruction; approximation; image dependent knowledge; image independent knowledge; adaptive method; feature set; feature space; self-diagnostic method; surface reconstruction model
Citation:
M.S. Lew, T.S. Huang, K. Wong, "Learning and Feature Selection in Stereo Matching," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 16, no. 9, pp. 869-881, Sept. 1994, doi:10.1109/34.310682
Usage of this product signifies your acceptance of the Terms of Use.