Model-based recognition of 3D objects from one view
In this work we treat major problems of object recognition which have previously received little attention. Among them are the loss of depth information in the projection from 3D to 2D, and the complexity of finding feature correspondences in general cases. This treatment enables us to recognize objects in difficult real-world situations.
It is well known that there are no geometric invariants of a projection from 3D to 2D. However, given some modeling assumptions about the 3D object, such invariants can be found. The modeling assumptions can be either a particular model or a generic assumption about a class of models. Here we deal with both situations. We find invariant connections between a 2D image and a 3D model under general projective projection. We give a geometric interpretation of the method as an invariant model in 3D invariant space, illuminated by invariant light rays, converging to an invariant camera center in the same space. We demonstrate the method on real images.
This work differs from related work in the following ways: 1) Most work in invariants has concentrated on transformations of the same dimensionality, mainly 2D to 2D projections. We deal here with the problem of projecting a 3D object onto a 2D image, which is of greater relevance to object recognition. 2) Much of the previous work is done on multiple images, such as the work on camera calibration. This usually requires knowledge of the correspondence between images. We concentrate on single images, but we also apply our method to finding correspondences in multiple images without prior knowledge or expensive search.
Keywordsobject recognition model based vision invariants
Unable to display preview. Download preview PDF.
- 1.E. Barrett. Personal communication.Google Scholar
- 2.P. Beardsley, P. Torr, and A. Zisserman. 3D Model Aquisition from Extended Image Sequences. Proc. European Conf. on Computer Vision, 683–695, 1996.Google Scholar
- 3.J. Ben-Arie, Z. Wang, and R. Rao. Iconic Recognition with Affine-Invariant Spectral Signatures. Proc. ICPR A, 672–676, 1996.Google Scholar
- 4.T.O. Binford and T.S. Levitt. Model-Based Recognition of Objects in Complex Scenes. Proc. DARPA Image Understanding Workshop, 89–100, 1996.Google Scholar
- 5.A. Bruckstein, E. Rivlin, and I. Weiss. Scale Space Invariants for Recognition. Machine Vision and Applications, 15:335–344, 1997.Google Scholar
- 6.J.B. Burns, R. Weiss, and E.M. Riseman. View Variation of Point Set and Line Segment Features. Proc. DARPA Image Understanding Workshop, 650–659, 1990.Google Scholar
- 8.R.W. Curwen and J.L. Mundy. Grouping Planar Projective Symmetries. Proc. DARPA Image Understanding Workshop, 595–605, 1997.Google Scholar
- 9.R. Deriche, Z. Zhang, Q.T. Luong, and O.D. Faugeras. Robust Recovery of the Epipolar Geometry for an Uncalibrated Stereo Rig. Proc. European Conf. on Computer Vision, A567–576, 1994.Google Scholar
- 10.H. Guggenheimer. Differential Geometry. Dover, New York, 1963.Google Scholar
- 11.J.P. Hopcroft, D.P. Huttenlocher, and P.C. Wayner. Affine Invariants for Model-Based Recognition. In Geometric Invariance in Machine Vision, eds. J.L. Mundy and A. Zisserman, MIT Press, Cambridge, MA, 1992.Google Scholar
- 12.D. Jacobs. Space Efficient 3D Model Indexing. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 439–444, 1992.Google Scholar
- 13.D. Jacobs and R. Basri. 3-D to 2-D Recognition with Regions. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 547–553, 1997.Google Scholar
- 14.D. Keren, R. Rivlin, I. Shimshoni, and I. Weiss. Recognizing 3D Objects using Tactile Sensing and Curve Invariants. University of Maryland Technical Report CS-TR-3812, 1997.Google Scholar
- 16.P. Meer and I. Weiss. Point/Line Correspondence under Projective Transformations. Proc. ICPR A, 399–402, 1992.Google Scholar
- 17.R. Mohr, L. Quan, and F. Veillon. Relative 3D Reconstruction using Multiple Uncalibrated Images. Intl. J. Robotics Research, 14:619–632, 1995.Google Scholar
- 18.S. Peleg, A. Shashua, D. Weinshall, M. Werman, and M. Irani. Multi-Sensor Representation of Extended Scenes using Multi-View Geometry. Proc. DARPA Image Understanding Workshop, 79–83, 1997.Google Scholar
- 21.C.A. Rothwell. Object Recognition through Invariant Indexing. Oxford University Press, Oxford, 1995.Google Scholar
- 24.G. Sparr. Projective Invariants for Affine Shapes of Point Configurations. Proc. 1st Workshop on Invariance, 151–170, 1991.Google Scholar
- 25.C.E. Springer. Geometry and Analysis of Projective Spaces. Freeman, San Francisco, 1994.Google Scholar
- 26.P.F. Stiller, C.A. Asmuth, and C.S. Wan. Invariant Indexing and Single View Recognition. Proc. DARPA Image Understanding Workshop, 1423–1428, 1994.Google Scholar
- 27.B. Strumpel. Algorithms in Invariant Theory. Springer Verlag, New York, 1993.Google Scholar
- 30.B. Vijayakumar, D.J. Kriegman, and J. Ponce. Structure and Motion of Curved 3D Objects from Monocular Silhouettes. Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 327–334, 1996.Google Scholar
- 31.I. Weiss. Geometric Invariants of Shapes. Proc. DARPA Image Understanding Workshop, 1125–1134, 1988.Google Scholar
- 35.I. Weiss. 3D Curve Reconstruction from Uncalibrated Cameras. University of Maryland CS-TR-3605. Also in Proc. ICPR A, 323–327, 1996.Google Scholar
- 36.I. Weiss. Model-based Recognition of 3D Curves from One View. Journal of Mathematical Imaging and Vision, in press, 1997.Google Scholar
- 37.M. Zerroug and R. Nevatia. Using Invariance and Quasi-invariance for the Segmentation and Recovery of Curved Objects. In Lecture Notes in Computer Science 825, Springer-Verlag, Berlin, 1994.Google Scholar