Under an assumption of monotonicity it was shown that

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: aper we investigated the feasibility of generating new views of a scene from two basis views. Under an assumption of monotonicity, it was shown that the problem is theoretically well-posed. This result is signi cant in light of the fact that it is not possible to fully recover the structure of the scene due to the aperture problem. Furthermore, we demonstrated that a particular range of views can be generated by linear interpolation of the basis images, if the basis images are rst recti ed. This result provides a theoretical basis for morphing techniques based on geometric image interpolation 1, 7, 8, 9, 10 and provides a simple way of generating new views of a scene. Finally, a scanline algorithm for interpolating two basis images was described that requires only a small number of userprovided feature correspondences. The application of the method was demonstrated on real images. 13 14 15 16 References 1 S. Toelg and T. Poggio, Towards an examplebased image compression architecture for video 7 conferencing," A.I. Memo No. 1494, M.I.T., Boston, MA, June 1994. L. McMillan and G. Bishop, Head-tracked stereoscopic display using image warping," in Proc. SPIE 2409A, 1995. R. Kumar, P. Anandan, and K. Hanna, Direct recovery of shape from multiple views: A parallax based approach," in Proc. ICPR, pp. 685 688, 1994. S. E. Chen and L. Williams, View interpolation for image synthesis," in Proc. SIGGRAPH 93, pp. 279 288, 1993. S. Ullman and R. Basri, Recognition by linear combinations of models," IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 13, no. 10, pp. 992 1006, 1991. S. Laveau and O. Faugeras, 3-D scene representation as a collection of images and fundamental matrices," Tech. Rep. 2205, INRIA, SophiaAntipolis, France, February 1994. G. Wolberg, Digital Image Warping. Los Alamitos, CA: IEEE Computer Society Press, 1990. T. Beier and S. Neely, Feature-based image metamorphosis," in Proc. SIGGRAPH 92, pp. 35 42, 1992. T. Poggio and R. Brunelli, A novel approach to graphics," A.I. Memo No. 1354, M.I.T., Boston, MA, February 1992. D. Beymer, A. Shashua, and T. Poggio, Example based image analysis and synthesis," A.I. Memo No. 1431, M.I.T., Boston, MA, November 1993. J. J. Koenderink and A. J. van Doorn, A ne structure from motion," Opt. Soc. Am. A, vol. 8, pp. 377 385, 1991. H. H. Baker and T. O. Binford, Depth from edge and intensity based stereo," in Proc. 7th International Joint Conference on Arti cial Intelligence, pp. 631 636, 1981. Y. Ohta and T. Kanade, Stereo by intra- and inter-scanline search using dynamic programming," IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 7, no. 2, pp. 139 154, 1985. T. Poggio, V. Torre, and C. Koch, Computational vision and regularization theory," Nature, vol. 317, pp. 314 319, 1985. S. M. Seitz and C. R. Dyer, Complete structure from four point correspondences," in Proc. Intl. Conf. on Computer Vision, 1995. To appear. C. Tomasi and T. Kanade, Shape and motion from image streams under orthography: a factorization method," Intl. Journal of Computer Vision, vol. 9, no. 2, pp. 137 154, 1992. Ι1 Ι 1.5 Ι2 ^ Ι1 ^ Ι 1.5 ^ Ι2 Figure 4: View interpolation control ow. The two original images are at top-left and top-right and an intermediate synthesized view is at top-center. The corresponding recti ed images are shown below the originals. The arrows show the ow of the algorithm, from original to recti ed to interpolated to derecti ed. Ι1 Ι 1.5 Ι2 Figure 5: Interpolation of images of a cube and a stapler. Original images are at left and right, and the interpolated image is in the center. Note that a metallic surface of the stapler in the left image is occluded in the right image, locally violating the assumption of monotonicity and causing local blurring in the interpolated image. Other local artifacts, such as an incorrect region near the top of the cube in I1:5 , result from errors in correspondence. 8...
View Full Document

Ask a homework question - tutors are online