D. P. Huttenlocher and S. Ullman. Object Recognition using Alignment. Proc. International Conference on Computer Vision, pages 102--111, 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....scenes. Preliminary modeling and recognition results are presented. 1 Introduction This paper addresses the problem of recognizing threedimensional (3D) objects in photographs. Traditional feature based geometric approaches to this problem, for example alignment and interpretation trees [6, 8], enumerate all triples of image features before pose consistency constraints can be used to confirm or discard competing match hypotheses. Appearance based techniques, on the other hand, use rich local descriptions of the image brightness pattern to select a relatively small set of promising ....

....its right and top side. In particular (and not surprisingly) a match between images of the same affine invariant patches contains exactly the same information as a match between triples of points. It is thus clear that all the machinery of structure from motion [3, 5, 23] and pose estimation [6, 8] from point matches can be exploited in our modeling and object recognition tasks. Reasoning in terms of multi view constraints associated with the matrix will provide in this paper a unified and convenient representation for all stages of both tasks, but one should always keep in mind the ....

[Article contains additional citation context not shown here]

D.P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. ICCV, pages 102--111, 1987.

....used to check the consistency of all potential matches of ribbons to object components. However, the actual calculation of bounds for such general constraints was mathematically difficult and approximations had to be used that did not lead to exact solutions for viewpoints. The alignment method [6] proposed by Huttenlocher and Ullman utilized minimal sets of features which suffice to establish a unique transformation between a model and its hypothesized instance in the image. For each match a corresponding transformation is computed, and the set of model edges is transformed to the image to ....

D.P. Huttenlocher and S. Ullman, "Object Recognition Using Alignment," Proc. of the 1'st Int. Conf. on Computer Vision, pp. 102-111, London, 1987.

....either have implemented a simple alignment algorithm or a simple Hough transform algorithm, we compare the quality of the results from optimal geometric matching to the quality of results obtained by implementations of those algorithms. For the implementation of geometric matching by alignment [21] used in this comparison, every possible pair of image features was put into correspondence with every possible pair of model features. For each such correspondence, the alignment 23 nclutter alignment branch andbound speedup 50 0.676 0.62 1.08 100 2.16 1.8 1.20 150 4.56 3.5 1.30 200 7.87 ....

....for high resolution matching problems; matchlist based methods are now routinely applied to matching geometric primitives and models in very large document images with subpixel accuracy. Exact Error Propagation in Correspondence Methods Correspondence based methods, meaning methods like alignment [21] and interpretation trees [16] can be extended with exact error propagation methods like those described in [1] Such methods will, given a set of correspondences between image and model features and given error bounds, predict reliable bounds on the possible locations of the 29 remaining model ....

[Article contains additional citation context not shown here]

D. P. Huttenlocher and S. Ullman. Object Recognition Using Alignment. In International Conference on Computer Vision, pages 102--111, London, England, June 1987. IEEE, Washington, DC. 34

....use analytic pose recovery methods because they only require the minimal number of feature correspondences. Perhaps the most famous pose recognition technique was proposed by Fischler and Bolles [28] in their seminal paper detailing RANSAC. This approach is also commonly referred to pose alignment [32]. The method of RANSAC begins with an hypothesis stage where the minimum number of image features needed to obtain a pose estimate are selected at random and assigned to arbitrary object correspondences. The pose is then calculated using these image and object correspondences and is used to ....

D. Huttenlocher and S. Ullman. Object recognition using alignment. In Proceedings, First International Conference on Computer Vision, pages 102--111, London, UK, 1987.

....of an object lie in a lower dimensional linear subspace and there can exist some algebraic constraints for recognition of objects in multiple views. A number of approaches have been proposed for planar shape recognition. Algorithms for planar shape recognition include recognition by alignment [5], polygonal approximation [9] based on geometrically invariant features [6] etc. Boundaries are also recognised by modelling the boundary in a transform domain like the Fourier one [14] All these algorithms limited their attention to similarity transformation between views. In most practical ....

D. P. Huttenlocher and S. Ullman. Object Recognition using Alignment. Proc. International Conference on Computer Vision, pages 102--111, 1987.

....scenes. Preliminary modeling and recognition results are presented. 1 Introduction This paper addresses the problem of recognizing threedimensional (3D) objects in photographs. Traditional feature based geometric approaches to this problem, for example alignment and interpretation trees [6, 8], enumerate all triples of image features before pose consistency constraints can be used to confirm or discard competing match hypotheses. Appearance based techniques, on the other hand, use rich local descriptions of the image brightness pattern to select a relatively small set of promising ....

....right and top side. In particular (and not surprisingly) a match between m 2 images of the same affine invariant patches contains exactly the same information as a match between m triples of points. It is thus clear that all the machinery of structure from motion [3, 5, 23] and pose estimation [6, 8] from point matches can be exploited in our modeling and object recognition tasks. Reasoning in terms of multi view constraints associated with the matrix S will provide in this paper a unified and convenient representation for all stages of both tasks, but one should always keep in mind the ....

[Article contains additional citation context not shown here]

D.P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. ICCV, pages 102--111, 1987.

.... as checking for model features in the order of the features prior probability [7, 12] in a coarse to fine hierarchy [6] or as predicted from the current scene interpretation [4, 8] Classic hypothesize and test paradigms have been demonstrated in RANSAClike [5, 19] and alignment techniques [9, 10]. The method proposed here is more similar to the latter in that testing of hypotheses is done with respect to the full data rather than on a sparse feature representation. It significantly di#ers from both, however, in that it recommends a certain order of hypotheses to be tested, which is indeed ....

Huttenlocher, D. P., and Ullman, S. Object recognition using alignment. In Proc. Intern. Conf. Comput. Vision (1987), pp. 102--111.

....correspondence. The transformation thus found is then used to project the rest of the model into the image to search for other corresponding features. The size of the search space is polynomial, O(n ) to be precise. This overall method has come to be associated with Huttenlocher and Ullman ([HU87]) who dubbed it alignment , though other previous work used transformation space search (for example, the Hough transform method [Bal81] as well as [Bai84, FB80, TM87] and others) One of the contributions of Huttenlocher s work was to show that a feature pairing of size 3 was necessary and ....

D.P. Huttenlocher and S. Ullman. Object Recognition Using Alignment. In 1st Int. Conf. Comp. Vis., pages 102--111, 1987.

....correspondence. The transformation thus found is then used to project the rest of the model into the image to search for other corresponding features. The size of the search space is polynomial, O(n ) to be precise. This overall method has come to be associated with Huttenlocher and Ullman ([HU87]) who dubbed it alignment , though other previous work used transformation space search (for example, the Hough transform method [Bal81] as well as [Bai84, FB80, TM87] and others) One of the contributions of Huttenlocher s work was to show that a feature pairing of size 3 was necessary and ....

D.P. Huttenlocher and S. Ullman. Object Recognition Using Alignment. In 1st Int. Conf. Comp. Vis., pages 102--111, 1987.

....A common approach to recognition uses features (suchaspoints or edges) to represent objects. An object is recognized in this approach if there exists a viewpointfrom which the model features coincide with the corresponding image features, e.g. Roberts, 1965, Fischler and Bolles, 1981,Lowe, 1985, Huttenlocher and Ullman, 1987, Basri and Ullman, 1988, Thompson and Mundy,1987, Ullman and Basri, 1991] Since images often are noisy and models occasionally are imperfect, it is rarely the case that a model aligns perfectly with the image. Systems therefore look for a model that reasonably aligns with the image. ....

....the other points and measuring their distance from the corresponding image points. Three [Fischler and Bolles, 1981,Rives et al. 1981, Haralick et al. 1991] or four [Horaud et al. 1989] points are required under perspective projection, and three points under weak perspective [Ullman, 1989, Huttenlocher and Ullman, 1987] The obtained distance critically depends on the choice of alignmentkey. Differentchoices produce different distance measures between the model and the image. The results almost always are sub optimal, since it is generally better to match all points with small errors than to exactly match a ....

Huttenlocher, D. P. and Ullman, S. (1987). Object recognition using alignment. In Proceedings of the 1st International Conference on Computer Vision, pages 102--111, London, England. IEEE, Washington, DC.

....of Applied Mathematics, Weizmann Institute of Science, Rehovot 76100, Israel Introduction Visual object recognition requires the identification of objects observed from different viewpoints. In recent years several attempts have been made to approach this problem using an alignment approach [5, 7, 9, 12, 17, 18]. In this paper we shall consider the recognition of rigid objects bounded by smooth surfaces, using an alignment approach. Alignment is a two stage process. Given a model object and an image object, in the first stage a transformation is sought, that would bring the model object to a position ....

....will not be discussed here. The transformation may be determined by a small set of corresponding features, identified in both the model and the image. For example, three non colinear points on the image, and their corresponding points on the model determine uniquely the transformation [8, 9, 12, 18]. Two points and a line or three lines may also serve for this purpose. 1.1 The Prediction Problem In this paper we address ourselves to the second stage of the alignment process. We present an approach for solving the following problem: Let = M,M2, M) be a set of object models. Let T be a ....

[Article contains additional citation context not shown here]

Huttenlocher, D.P. & Ullman, S., 1987. Object recognition using alignment. Proc. of IOOV Conf. (London) 102-111.

No context found.

D. P. Huttenlocher and S. Ullman. Object Recognition using Alignment. Proc. International Conference on Computer Vision, pages 102--111, 1987.

No context found.

D.P. Huttenlocher, S. Ullman, Object recognition using alignment, Proceedings of International Conference on Computer Vision (1987) 102 -- 111.

No context found.

D.P. Huttenlocher and S. Ullman, "Object recognition using alignment," in Proc. 1st Int. Conf. on Computer Vision, London, Los Alamitos CA, 1987, pp. 102--1112, IEEE Computer Society Press.

No context found.

Huttenlocher, D. & Ullman, S. (1987), Object recognition using alignment, in `Proceedings, 1ST International Conference on Computer Vision', London, UK., pp. 102--111.

No context found.

D. Huttenlocher and S. Ullman. Object recognition using alignment. In Proceedings, First International Conference on Computer Vision, pages 102--111, London, UK, 1987.

No context found.

D. P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. 1st Internat. Conf. Comput. Vision, pages 102-111, 1987.

No context found.

D.P. Huttenlocher and S. Ullman, "Object recognition using alignment," in Proc. 1st Int. Conf. on Computer Vision, London, Los Alamitos CA, 1987, pp. 102--1112, IEEE Computer Society Press.

No context found.

D. P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. ICCV, pages 102--111, 1987.

No context found.

D. Huttenlocher and S Ullman, "Object recognition using alignment", Proceeding of international conference on computer vision, London, pages 102-111,1987

No context found.

D.P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. Int. Conf. Comp. Vision, pages 102--111, London, U.K., June 1987.

No context found.

D.P. Huttenlocher and S. Ullman. Object recognition using alignment. In Proc. of ICCV, pages 7278, 1987.

No context found.

D.P. Huttenlocher and S. Ullman, "Object Recognition Using Alignment," Proc. of the 1'st Int. Conf. on Computer Vision, pp. 102-111, London, 1987.

No context found.

Huttenlocher D. and Ullman S., \Object recognition using alignment", Proc. ICCV (London), 102-111, 1987.

No context found.

D. I-Iuttenlocher and S. Ullman, "Object recognition using alignment," First Int. Conf. Computer Vision, London, 1987, pp.102-111.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC