R. Vaillant. Using occluding contours for 3D object modeling. In European Conference on Computer Vision, pages 454--464, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....by a set of 2D segments extracted from the initial image acquired by the camera. We assume that these segments correspond to either a limb of a cylinder, either a 3D segment. As the structure estimation method is peculiar to each kind of primitive, a recognition process is necessary. Vaillant [14] proposed a criterion to determine whether the considered 2D segment corresponds to the projection of a 3D segment or to a cylinder limb. The observed segment is assumed to be part of a cylinder, and according to this hypothesis the radius of the cylinder is estimated. Since a 3D segment is ....

....2D segment corresponds to the projection of a 3D segment or to a cylinder limb. The observed segment is assumed to be part of a cylinder, and according to this hypothesis the radius of the cylinder is estimated. Since a 3D segment is characterized by a zero radius, the criterion proposed in [14] is based on the probability for zero to be in the interval of confidence: r 2er ( 0 ( r. Our experiments have shown that this criterion is not always able to correctly detect which kind of primitive was observed. So, in order to obtain a robust criterion, we have developed a method based on a ....

VAILLANT R. Using occluding contours for 3D objects modeling. Proc of ECCV, Antibes, France, April 1990.

....with subtracted background. This is an automatic but low level method and thus relatively easy to implement but not very robust in practice. Automated silhouette edge detectors have been developed and could be implemented FIG. 8. Automatically tracked silhouette of a walking person. for this use [33]. In this work, we have used semi automated tools to allow the user to quickly sketch the silhouette edges [25] We also experimented with a snake based silhouette tracker that uses 3 D point trajectory information. The silhouette is interactively initialized in the first frame, as shown in ....

R. Vaillant and O.D. Faugeras. Using Occluding Contours for 3D Object Modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, February 1992.

....Many authors cite the use of Canny edge detectors in images with subtracted background. This is an automatic but low level method and thus relatively easy to implement but not very robust in practice. Automated silhouette edge detectors have been developed and could be implemented for this use [33]. In this work, we have used semi automated tools to allow the user to quickly sketch the silhouette edges [25] We also experimentedwith a snake based silhouette tracker that uses 3 D point trajectory information. The silhouette is interactively initialized in the first frame, as shown in ....

R. Vaillant and O.D. Faugeras. Using Occluding Contours for 3D Object Modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, February 1992.

....profiles. Those works were based on a continuous representation of the objects among a set of projections. Some concentrates on finding hidden surfaces and detecting occlusions between the objects, like [16] others tried to reconstruct surfaces from the set of different camera views ( 4] 14] 8] [18], 1] 6] Their main problem was to determine very 3 geometric transform Visualization Diagnostic Surface reconstruction Edge detection Measurement, surgical planning, prosthesis design, Dual to object space Oriented dual curves Edge detection Surfaces Sinogram Scanner Geometric ....

....Gauss and mean curvature. They also point out that the profile of a non convex surface generates singularities, such as swallow tails , where the cusps corresponds to the inflection points of the original curve. Another important work has been performed by Regis Vaillant and Olivier Faugeras in [18]. They derive a local geometric transform from the envelope theorem which allow them to reconstruct 3D surfaces from 2D profiles. Their goal was to classify the different kinds of edges into the images (edge, wedge, profile) and to find back the shape and the differential properties of simple ....

[Article contains additional citation context not shown here]

R. Vaillant and O. Faugeras. Using occluding contours for 3d object modeling. International Advanced Robotics Program, october 1989. 34

....points can be determined, i.e. one obtains a surface patch rather than a set of points. The problem of reconstructing a smooth surface from its profiles has been explored for known planar motion by Giblin and Weiss [GW87] and subsequently for more general known motion by Vaillant and Faugeras [Vai90, VF92] and Cipolla and Blake [BC90, CB90, CB92] These approaches are either based on a differential formulation and analysis, or they use curve fitting but still only use three frames. First order temporal derivatives are usually computed as differences from pairs of frames, and second order ....

....smoothing along the curves to reduce noise. This can be viewed as part of the edge extraction stage. that line. Under orthographic projection, if the camera motion is planar, then all of the epipolar curves will be planar as well. Cipolla and Blake [CB90, CB92] and Vaillant and Faugeras [Vai90, VF92] noticed that to compute the curvature of a planar curve from three tangent rays, one can determine a circle which is tangent to these rays. The assumption that one needs to make is that the surface remains on the same side of the tangent rays. This is true for intervals of the curve which do not ....

R. Vaillant. Using occluding contours for 3D object modeling. In First European Conference on Computer Vision (ECCV'90), pages 454--464, Antibes, France, April 23--27 1990. Springer-Verlag.

....and three views. 1.1 Related literature on curve matching Curve matching is a reoccurring topic in the stereo and motion literature, with the main application being the recovery of 3D geometry. A subsidiary goal has been to qualitatively distinguish between surface curves and apparent contours [2, 25]. Curves are a natural next step after lines, since they are a more faithful representation of a contour which would otherwise be represented by a polygonal chain. In this paper we are not considering closely spaced views where snake tracking is viable. Basic criteria for curve matching in stereo ....

R. Vaillant. Using occluding contours for 3D object modeling. In Proc. ECCV, pages 454--464, 1990.

....by a set of 2D segments extracted from the initial image acquired by the camera. We assume that these segments correspond to either a limb of a cylinder, either a 3D segment. As the structure estimation method is peculiar to each kind of primitive, a recognition process is necessary. Vaillant [14] proposed a criterion to determine whether the considered 2D segment corresponds to the projection of a 3D segment or to a cylinder limb. The observed segment is assumed to be part of a cylinder, and according to this hypothesis the radius of the cylinder is estimated. Since a 3D segment is ....

....2D segment corresponds to the projection of a 3D segment or to a cylinder limb. The observed segment is assumed to be part of a cylinder, and according to this hypothesis the radius of the cylinder is estimated. Since a 3D segment is characterized by a zero radius, the criterion proposed in [14] is based on the probability for zero to be in the interval of confidence: r Gamma 2oe 0 r. Our experiments have shown that this criterion is not always able to correctly detect which kind of primitive was observed. So, in order to obtain a robust criterion, we have developed a method based ....

VAILLANT R. Using occluding contours for 3D objects modeling. Proc of ECCV, Antibes, France, April 1990.

....[Szeliski, 1990] This bounding volume can provide a non linear (inequality) constraint on the position of surface points. Tracking the silhouettes through three or more images can also be used to estimate the location and curvature of points on the limb of the object [Giblin and Weiss, 1987; Vaillant, 1990; Cipolla and Blake, 1990] Combining silhouette based and flow based approaches should yield an algorithm that works for a much wider variety of object shapes and textures. Our shape from rotation algorithm would be even more useful if we could change the position of the camera and or the object. ....

R. Vaillant. Using occluding contours for 3D object modeling. In First European Conference on Computer Vision (ECCV'90), pages 454--464, SpringerVerlag, Antibes, France, April 23--27 1990.

....determined, i.e. one obtains a surface patch rather than a set of points. The problem of reconstructing a smooth surface from its profiles has been explored for known planar motion by Giblin and Weiss [Giblin and Weiss, 1987] and subsequently for more general known motion by Vaillant and Faugeras [Vaillant, 1990; Vaillant and Faugeras, 1992] and Cipolla and Blake [Blake and Cipolla, 1990; Cipolla and Blake, 1990; Cipolla and Blake, 1992] These approaches are either based on a differential formulation and analysis, or they use curve fitting but still only use three frames. First order temporal ....

....is linear, then the epipolar planes form a pencil of planes containing that line. Under orthographic projection, if the camera motion is planar, then all of the epipolar curves will be planar as well. Cipolla and Blake [Cipolla and Blake, 1990; Cipolla and Blake, 1992] and Vaillant and Faugeras [Vaillant, 1990; Vaillant and Faugeras, 1992] noticed that to compute the curvature of a planar curve from three tangent rays, one can determine a circle which is tangent to these rays. See Figure 2. The assumption that one needs to make is that the surface remains on the same side of the tangent rays. This is ....

R. Vaillant. Using occluding contours for 3D object modeling. In First European Conference on Computer Vision (ECCV'90), pages 454--464, Springer-Verlag, Antibes, France, April 23--27 1990.

....and stereo terms presented in Section 2.4. The results are shown in Figure 15(c) d) and (e) The silhouettes used in the two examples above have been entered semiautomatically. But here again, we could take advantage of automatically extracted ones [ Cipolla and Blake, 1990, Liedtke et al. 1991, Vaillant and Faugeras, 1992 ] 4 Conclusion We have presented a surface reconstruction method that uses an object centered representation to recover 3 D surfaces. Our method uses both monocular shading cues and stereoscopic cues from any number of images while correctly handling self occlusions. It can also take advantage ....

R. Vaillant and O.D. Faugeras. Using Occluding Contours for 3D Object Modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, February 1992.

....of the black cubes in its path. In the general case, this occupancy constraint is complicated to represent and use. A more practical approach is to track the silhouette over three or more frames to estimate the location and curvature of points on the surface of the object [Giblin and Weiss, 1987; Vaillant, 1990] 10 Discussion 23 10 Discussion Given the current performance of our algorithm, is it fair to call it a real time method The microwave turntable that we use to rotate our object turns at 0.5 revolutions per minute. The time to acquire, process, and display a 512 Theta 480 image on our ....

R. Vaillant. Using occluding contours for 3D object modeling. In First European Conference on Computer Vision (ECCV'90), pages 454--464, SpringerVerlag, Antibes, France, April 23--27 1990.

....[16] pointed out that a surface can be reconstructed from the envelope of all its tangent planes, which are computed directly from the occluding contours and the motion of the viewer. This was really achieved with the works of Arbogast Mohr [1] Cipolla Blake [9, 3, 10] and Vaillant Faugeras [32, 31, 33]; The work of Vaillant et al. presented an original method for detecting and reconstructing the occluding contours, using a sequence of at least three images. Arbogast et al. and Cipolla et al. considered the computation of the local surface curvature and the surface reconstruction from the ....

R. Vaillant. Using occluding contours for 3D object modeling. In O. Faugeras, editor, Proceedings of the 1st European Conference on Computer Vision, Antibes, France, pages 454--464. Springer-Verlag, April 1990.

No context found.

R. Vaillant. Using occluding contours for 3D object modeling. In European Conference on Computer Vision, pages 454--464, 1990.

No context found.

Vaillant, R., Using Occluding Contours for 3D Object Modeling, Proc. ECCV1, Springer-Verlag, 1990.

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