Faugeras, O., Lustman, F., and Toscani, G., "Motion and Structure from Motion from Point and Line Matches", Proc. 1st Intl. Conf. on Computer Vision, p25-34, 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of the filter saturates as the number of visible points increases beyond 20. The performance also degrades as the points move far away from the viewer and as the structure approaches a plane. Under these conditions, in fact, the matrix approaches rank 6, rather than its normal rank of 8 [24, 26]. We have tested the essential filter in local coordinates, both implemented using the IEKF and the two D iteration described in appendix A, and the essential filter in the embedding space. We now comment on the performance of each filter on the reference simulation experiment, highlighting some ....

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. Proc. of the IEEE Conf. ICCV, 1987.

.... [6, 9, 10] as well as displacement fields [8, 12, 14] have formed the foundation of most research on rigid motion analysis that addressed the 3D motion problem by first approximating the motion field through the optical flow and then interpreting the optical flow to obtain 3D motion and structure [2, 7, 13, 15]. The difficulties involved in the estimation of optical flow have recently given rise to a small number of studies considering as input to the visual motion interpretation process some partial optical flow information. In particular the projection of the optical flow on the gradient direction, ....

O. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In Proc. International Conference on Computer Vision, pages 25--34, 1987.

....when the camera field of view is small. To improve the estimation of the epipolar geometry, outlier detection and elimination have also been attempted [42] When point like features cannot be reliably extracted from image pairs, camera self calibration using linear features has also been examined [11, 33, 41, 22, 8, 29, 37]. If a mobile camera is mounted on a platform such that the motion of the camera can be accurately controlled, the intrinsic parameters of the camera can be recovered independently of the extrinsic parameters [6, 2] The simultaneous recovery of both intrinsic (interior orientation) and extrinsic ....

Faugeras, O. D., Lustman, F., and Maybank, S. J. Motion and Structure from Motion from Point and Line Matches. In Proc. International Conference on Computer Vision (1987), pp. 24--34.

.... Numerous methods have been developed for motion recovery from image sequences, among them are algorithms that compute the motion directly from the grey level values or local measures of them [14, 20, 16, 3, 17] A second class of algorithms use feature points or optical flow to recover motion [1, 9, 15]. A probabilistic error minimization algorithm [26] can be used to recover motion in the presence of outliers. Another class of algorithms use explicit probability distributions of the motion vectors to calculate motion models [23] Black and Anandan presented [7] a framework for robust ....

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matching. In Int. Conf. on Computer Vision, pages 25--34, 1987.

....the environment, taking an image sequence as the input. Most of the time, the process of ego motion estimation starts by computing image motion and then using those vectors and properties of the motion field equations to estimate ego motion. Usual approaches are based on point correspondences [1], optical flow [2] or the so called direct methods [3, 4] For the usual cameras that can be modeled as the perspective projection of the 3D structure onto an image plane, ego motion is difficult to estimate. Even though the information about the observer motion is present in the images, it is ....

....are quite small, thus validating our approach. In the final experiment, the robot moves with a pure rotation. As mentioned before, for this special case when translation is zero T (0, 0, 0) Equation (9) cannot be applied. A least square estimate for# is obtained using the original odometry T = [ 1 0 0] T = 0.9990 0.0425 0.0109] 0 0 0] 0.0486 Table 1. Ego motion values and errors for pure translation. 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 ....

F. Lustman, O. Faugeras, and G. Toscani, "Motion and structure from motion from point and line matches," Proc. of First Int'l Conf. Computer Vision, June 1987.

....constraints are usually added to the motion model or to the scene structure. 3D motion is often estimated from the optical or normal flow derived between two frames [1] 11] 23] or from the correspondence of distinguished features (points, lines, contours) extracted from successive frames [24], 12] 8] Both approaches depend on the accuracy of the feature detection, which can not always be assured. Methods for computing the ego motion directly from image intensities were also suggested [10] 13] Camera rotations and translations can induce similar image motions [2] 9] causing ....

F. Lustman, O.D. Faugeras, and G. Toscani, "Motion and Structure from Motion from Point and Line Matching," Proc. First Int'l Conf. Computer Vision, pp. 25--34, London, 1987.

.... Numerous methods have been developed for motion recovery from image sequences; among them are algorithms that compute the motion directly from the grey level values or local measures of them [3, 14, 16, 17, 20] A second class of algorithms use feature points or optical flow to recover motion [1, 9, 15]. A probabilistic error minimization algorithm [26] can be used to recover motion in the presence of outliers. Another class of algorithms use explicit probability distributions of the motion vectors to calculate motion models [23] Black and Anandan presented [7] a framework for robust ....

O. D. Faugeras, F. Lustman, and G. Toscani, Motion and structure from motion from point and line matching, in Int. Conf. on Computer Vision, 1987, pp. 25--34.

....of the resulting functions. Furthermore, there exist techniques that first estimate rotation and, on the basis of the result subsequently estimate translation [9, 37, 38] techniques that do the opposite [1, 23, 26, 31, 33, 39, 42] and techniques that estimate all motion parameters simultaneously [7, 8, 17, 36, 45]. The positive depth constraint, which has been used for normal flow fields, is relatively new and is employed in the so called direct algorithms [8, 21, 27] One has to search for the 3D motion that is consistent with the input and produces the minimum amount of negative depth. Put differently, ....

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In Proc. International Conference on Computer Vision, pages 25--34, 1987.

....Algorithms for this problem of 3D interpretation from monocular motion can be broadly divided into two categories two frame and multi frame. Two frame algorithms first compute the relative orientation the translation and rotation between the camera positions at two time instants [1, 12, 14, 29]. Then the relative orientation is used to compute the 3D location for each imaged feature. In addition to advantages and disadvantages specific to instances of these algorithms, the two frame methods suffer from two major problems. First, for some motions, there are inherent ambiguities in the ....

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In IEEE First International Conference on Computer Vision, pages 25--34, 1987.

....UAV is a special case of the general one: All the image points correspond to coplanar points on the landing pad. It is well known that the case where all features points in the scene are coplanar is a degenerate case that makes the 8 point algorithm ill conditioned, giving poor estimation results [4]. Hence one needs algorithms specific to the planar case. The discrete version of the planar ego motion problem has been studied extensively [3, 12, 23] Here we only formulate the problem and briefly revisit well known results that can be found in [23] Our contribution is to the differential ....

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In IEEE International Conference on Computer Vision, 1987.

....many vision based applications. Numerous methods have been developed for motion recovery from image sequences, among them are algorithms that compute the motion directly from the grey level or general local measures [7, 11, 9, 2] A second class of algorithms use feature points to recover motion [4, 8]. A probabilistic error minimization algorithm [15] can be used to recover motion in the presence of outliers. Another class of algorithms use explicit probability distribution of the motion vectors to calculate motion models [13] Most of the methods cited above have problems when computing ....

O. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matching. In Int. Conf. on Computer Vision, pages 25--34, 1987.

.... and surface markings, many techniques have been developed for recovering the 3D contour locations from two or more images under known camera motion [MP79, MF80, Arn83, BBM87, BB89, MKS89] Techniques have also been developed for simultaneously estimating contour locations and camera positions [TH84, FLT87, Hor90]. However, for smooth curved surfaces, the critical set which generates the profile is different for each view. Thus, the triangulation applied in two frame stereo will not be correct along the occluding contour for smooth surfaces. For the same reason, it is often not possible to determine the ....

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In First International Conference on Computer Vision (ICCV'87), pages 25--34, London, England, June 1987. IEEE Computer Society Press.

....began to use primitives other than points. Such primitives include planar curves and straight lines (infinitely long lines, not line segments) For the case of straight lines the analysis proceeded similarly to the study of point correspondences. First, iterative algorithms [Liu and Huang, 1988; Faugeras et al. 1987] were derived, and later a linearization technique established the uniqueness properties of the approach [Spetsakis and Aloimonos, 1990] Not much research has been devoted to the extraction of motion from planar contours. Bergholm [1988] studied uniqueness properties. In general it is ....

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In Proc. International Conference on Computer Vision, pages 25--34, 1987.

....[10] in the general case and the algorithm proposed by Tsai and Huang [16] in the planar case have popularized the essential matrix representation. However, the original algorithms are reported to be very sensitive to noise. More robust algorithms have been developed since, and can be found in [2, 18]. Under orthogonal projections, Ullman [17] shows that three views of four points in a nonplanar configuration uniquely determine their structure. Although at least three distinct views are necessary under orthogonal projections, Koenderink and van Doorn [8] point out that the structure of an ....

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In Proc. First Int'l Conf. Comput. Vision, pages 25--34, London, UK, 1987.

....any constraints upon the camera displacements. Most of the work has been to write down the constraints that come out of the observation of one line in three views and to propose algorithms for solving efficiently the system of nonlinear equations resulting from the observation of p such lines [21, 10, 22, 23, 20, 39, 45, 47]. The question of the critical sets of lines has only been brought up recently by Thomas Buchanan [4, 5] The problem can be phrased in a way that is similar to that of points: for a given algorithm, does there exist sets of lines such that, no matter how many lines we observe in these sets, we ....

....families of lines, called line complexes, which are essential in the analysis of previously published algorithms on the estimation of camera displacement from line correspondences. In section 4 we use these complexes for characterizing the critical sets Phi of previously published algorithms [21, 10, 22, 23]. In section 5 we provide an alternative description of Buchanan s critical set of lines Psi which sheds more light on its algebraic structure and on its relationship with the basic equations which govern the estimation of the camera displacement. Section 6 provides graphical descriptions of some ....

[Article contains additional citation context not shown here]

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In Proc. First Int'l Conf. Comput. Vision, pages 25--34, London, UK, 1987.

....(i.e. x 1 0 and x 2 0) Self Calibration and Metric Reconstruction of an Uncalibrated Stereo 17 Once we have computed ff u and ff v , we can compute the essential matrix by E = A T FA. The rotation and translation are then computed from E by any standard methods such as those described in [6, 30, 4]. We thus compute ff ul , ff vl , R l and t l for the left camera from the fundamental matrix F ll . In the same way, we can compute ff ur , ff vr , R r and t r for the right camera from the fundamental matrix F rr . It remains to compute the rotation R s and the translation t s between the left ....

O. Faugeras, F. Lustman, and G. Toscani, "Motion and structure from motion from point and line matches," in Proc. First Int'l Conf. Comput. Vision, (London, UK), pp. 25--34, 1987.

No context found.

Faugeras, O., Lustman, F., and Toscani, G., "Motion and Structure from Motion from Point and Line Matches", Proc. 1st Intl. Conf. on Computer Vision, p25-34, 1987.

No context found.

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In First Int. Conf on Computer Vision, pages 25--34, WashingtonDC 1987.

No context found.

F. Lustman O.D. Faugeras and G. Toscani. Motion and structure from motion from point and line matching. In Proc. 1st International Conference on Computer Vision, pages 25--34, London, 1987.

No context found.

O.D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In IEEE International Conference on Computer Vision, 1987.

No context found.

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In First International Conference on Computer Vision (ICCV'87), pages 25--34, London, England, June 1987. IEEE Computer Society Press. 179

No context found.

O. D. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. In First Int. Conf on Computer Vision, pages 25--34, WashingtonDC 1987.

No context found.

O. Faugeras, F. Lustman, G. Toscani, \Motion and Structure from Motion from Point and Line Matches", Proc. 1st Int. Conf. on Computer Vision, pp. 25-34, 1987.

No context found.

O. Faugeras, F. Lustman, and G. Toscani. Motion and structure from motion from point and line matches. International Conference on Computer Vision, pages 2534, 1987.

No context found.

O.Faugeras, Lustman, Toscani, "Motion and Structure from Motion from Point and Lines Matches", Proceedings IEEE, pp. 25-34, 1987.

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