Abstract
A new approach is introduced to estimating object surfaces in three-dimensional space from a sequence of images. A 3D surface of interest here is modeled as a function known up to the values of a few parameters. Surface estimation is then treated as the general problem of maximum-likelihood parameter estimation based on two or more functionally related data sets. In our case, these data sets constitute a sequence of images taken at different locations and orientations. Experiments are run to illustrate the various advantages of using as many images as possible in the estimation and of distributing camera positions from first to last over as large a baseline as possible. In order to extract all the usable information from the sequence of images, all the images should be available simultaneously for the parameter estimation. We introduce the use of asymptotic Bayesian approximations in order to summarize the useful information in a sequence of images, thereby drastically reducing both the storage and the amount of processing required. This leads to a sequential Bayesian estimator for the surface parameters, where the information extracted from previous images is summarized in a quadratic form. The attractiveness of our approach is that now all the usual tools of statistical signal analysis, for example, statistical decision theory for object recognition, can be brought to bear; the information extraction appears to be robust and computationally reasonable; the concepts are geometric and simple; and essentially optimal accuracy should result. Experimental results are shown for extending this approach in two ways. One is to model a highly variable surface as a collection of small patches jointly constituting a stochastic process (e.g., a Markov random field) and to reconstruct this surface using maximum a posteriori probability (MAP) estimation. The other is to cluster together those patches constituting the same primitive object through the use of MAP segmentation. This provides a simultaneous estimation and segmentation of a surface into primitive constituent surfaces.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Ayache N., Faugeras O.D. 1987. Maintaining representations of the environment of a mobil robot. Intern. Symp. Robotics Res. 4, MIT Press; also INRIA Technical report, France.
Amblard, F.G., Cooper, D.B., Cernuschi-Frias, B. 1986. Estimation by multiple views of outdoor terrain modeled by stochastic processes. Proc. SPIE Cambridge Symp. Optical Optoelectronic Engineer.: Intelligent Robots and Computer Vision. Cambridge, MA, Oct. 28–31.
Barnard S., and Fishler M. 1982. Computational stereo. ACM Computing Surveys 14(4): 553–572.
Besl P.J., and Jain R.C. 1986. Invariant surface characteristics for 3D object reconstruction in range images. Comput. Vision, Graph., Image Process. 33: 33–80.
Bolle R.M., and Cooper D.B. 1986. On optimally combining pieces of information, with application to estimating 3D complex-object position from range data. IEEE Trans. PAMI, 8(5): 619–638.
Bolles R.C., Baker H.H., and Marimont D.H. 1987. Epipolar-plane image analysis: an approach to determining structure from motion. Intern. J. Comput. Vision 1(1): 7–55.
Boult, T.E., and Chen, L.H. 1988. Analysis of two new stereo algorithms. Proc. 1988 IEEE Conf. Comput. Vision Patt. Recog. Ann Arbor, 177–182.
Broida, T.J., and Chellappa, R. 1986. Kinematics and structure of a rigid object from a sequence of noisy images. Proc. IEEE Workshop on Motion: Representation and Analysis, Charleston, S.C., May, pp. 95–100.
Cernuschi-Frias, B., Belhumeur, P.N., and Cooper, D.B. 1985. Estimating and recognizing parameterized 3D objects using a moving camera. Proc. IEEE Comp. Soc. Conf. on Computer Vision and Pattern Recognition, San Francisco, June, pp. 167–171.
Cernuschi-Frias B., Cooper D.B., Hung Y.P., and Belhumeur P.N. 1989. Toward a model-based Bayesian theory for estimating and recognizing parameterized 3D objects using two or more images taken from different positions. IEEE Trans. PAMI 11: 1028–1052, Oct.
Cohen, F.S., and Cooper, D.B. 1983. Real-time textured image segmentation based on non-casual Markovian random field models. Proc. 3rd Intern. Conf. Robot Vision Sensory Control, Cambridge, MA, Nov. 449: 17–28.
Cohen F.S., and Cooper D.B. 1987. Simple parallel hierarchical and relaxation algorithms for segmenting non-casual Markovian random fields. IEEE Trans. PAMI 9(2): 195–219.
Cooper, D.B., Hung, Y.P., and Taubin, G. 1988. A new model-based stereo approach for 3D surface reconstruction using contours on the surface pattern. Proc. 2nd IEEE Intern. Conf. Comput. Vision, Tampa, FL, December, pp. 74–83.
Dereniak E.L., and Crowe D.G. 1984. Optical Radiation Detectors. Wiley: New York.
Derin H., and Elliott H. 1987. Modeling and segmentation of noisy and textured images using Gibbs random fields. IEEE Trans. PAMI 9(1): 39–55.
Duda D., and Hart P. 1973. Pattern Classification and Scene Analysis. Wiley: New York, p. 381.
Faugeras, O.D., Ayache, N., and Faverjon, B. 1986. Building visual maps by combining noisy stereo measurements. Proc. IEEE Intern. Conf. Robotics Automation, San Francisco, April 7–10, pp. 1433–1438.
Faugeras, O.D., Lustman, F., and Toscani, G. 1987. Motion and structure from motion from point and line matches. Proc. 1st Intern. Conf. Comput. Vision, London, pp. 25–34.
Geman S., and Geman D. 1984. Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images. IEEE Trans. PAMI, 6(6): 721–741.
Grimson W.E.L. 1981. From Images to Surfaces: A Computational Study of the Human Early Vision System, MIT Press: Cambridge, MA.
Hoff, W., and Ahuja, N. 1987. Extracting surfaces from stereo images: An integrated approach. Proc. 1st Intern. Conf. Comput. Vision, London, pp. 284–294.
Horn B.K.P. 1986. Robot Vision. MIT Press: Cambridge, MA.
Hung, Y.P. 1990. Three dimensional surface reconstruction using a moving camera: A model-based probabilistic approach. Ph.D. Thesis, Brown University; also, LEMS Tech. Report 63, Division of Engineering, Brown University, November 1989.
Hung, Y.P., and Cooper, David B. 1990. Maximum a posteriori probability 3D surface reconstruction using multiple intensity images directly. To appear in Proc. SPIE Conf. Sensing Reconstr. 3D Objects Scenes, February.
Kass M. 1988. Linear image features in stereopsis. Intern. J. Comput. Vision 1(4): 357–368.
Marr D., and Poggio T. 1979. A computational theory of human stereo vision. Proc. Roy. Soc. London, B204: 301–328.
Matthies, L., Szeliski, R., and Kanade, T. 1988. Incremental estimation of dense depth maps from image sequences. Proc. 1988 IEEE Conf. Comput. Vision Patt. Recog. Ann Arbor, 366–374.
Medioni G., and Nevatia R. 1985. Segment-based stereo matching. Comput. Vision, Graph., Image Process. 31: 2–18, July.
Ohta Y., and Kanade T. 1985. Stereo by intra- and inter-scanline search using dynamic programming. IEEE Trans PAMI 7(2): 139–154.
Subrahmonia, J., Hung, Y.P., and Cooper, D.B. 1989. Model-based segmentation and estimation of 3D surfaces from two or more intensity images using Markov random fields. LEMS Tech. Report 64, Division of Engineering, Brown University, November; also, under review for publication.
Szeliski, R. 1988. Estimating motion from sparse range data without correspondence. Proc. 2nd Intern. Conf. Comput. Vision, Tampa, FL, pp. 207–216.
Xu, G., Tsuji, S., and Asada, M. 1985. Coarse-to-fine control strategy for matching motion stereo pairs, Proc. 9th Intern. Joint Conf. Artif. Intell., Los Angeles, pp. 892–894.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Hung, YP., Cooper, D.B. & Cernuschi-Frias, B. Asymptotic bayesian surface estimation using an image sequence. Int J Comput Vision 6, 105–132 (1991). https://doi.org/10.1007/BF00128152
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF00128152