M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela. Direct methods for self-calibration of a moving stereo head. In Proc. 4th European Conference on Computer Vision, LNCS 1065, Cambridge, pages II:415--426, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....using three images, the intrinsic calibration of a camera undergoing planar motion can be recovered up to a two fold ambiguity. Due to the widespread use of binocular heads in robotics, the problem of autocalibrating a stereo rig from a sequence of images has also received considerable attention [55, 26, 6, 27, 25]. Owing to the redundancy of the information contained in a sequence of stereo images, such methods have the potential of being considerably more robust compared to self calibration methods employing monocular image sequences. Recently, attempts to solve the self calibration problem in the case ....

M. J. Brooks, L. de Agapito, D. Q. Huyng, and L. Baumela. Direct methods for self-calibration of a moving stereo head. In Fourth European Conference on Computer Vision, volume II, pages 415--426, April 1996.

....technique on grey level images enhances the tracking performance to a rate of 38 images per second. DRAFT PLAN EXECUTION BASED ON ACTIVE PERCEPTION: ADDING HINTS TO PLANS 10 B. 1 Stereo vision system The Stereo vision system determines the 3 D location of the beacon using a calibrated stereo head (Brooks, de Agapito, Huynh and Baumela 1996, Faugeras 1993) Two Detect behaviors, each attached to the left and right color images, are used in order to find the corresponding projections of the beacon in both images. Stereo measurements of the beacon location are gathered in fixed time intervals in which the control actions of the NMBs ....

Brooks, M. J., de Agapito, L., Huynh, D. Q. and Baumela, L.: 1996, Direct Methods for Self-Calibration of a Moving Stereo Head, Proceedings ECCV'96, Vol. 2, Cambridge, UK, pp. 415--426.

....case of a moving camera with variable and unknown focal length, but whose other intrinsic parameters are known. A practical self calibration algorithm was proposed by Azarbayejani and Pentland [1] Algorithms and closed form solutions for the two view case were given by Hartley and Brooks et al. [7, 3, 4, 10]. Newsam et al. derived the critical motions for the two view case [10] In this paper, we derive a complete characterization of critical motion sequences, for any number of views, and the critical motions for stereo systems. The paper is organized as follows. In x2 we provide some theoretical ....

....that (self ) calibration has a unique solution. 6.3 Planar Motion By planar motion we understand that both optical center and optical axis are bound to move in a single plane. It has been shown by Brooks et al. that for two views, this is always a critical configuration for self calibration [3]. If the two optical axes are parallel, then the configuration is critical according to both cases 1 and 3. If not, then the configuration is critical according to case 3. In that case, we can describe the configuration as that of a moving camera, that fixates a finite point (the intersection of ....

M.J. Brooks, L. de Agapito, D.Q. Huynh and L. Baumela, "Direct Methods for Self-Calibration of a Moving Stereo Head," ECCV, UK, 415-426, 1996.

....robust reconstructions to be computed either from an image pair obtained from static stereo head, or from two pairs of images obtained from a stereo head undergoing ground plane motion. Our work extends considerably the explicit analysis of the epipolar geometry presented in Brooks et al. [3]. Research most closely related to that presented here is the self calibration dynamic stereo work of Zhang et al. 25] and the special techniques for computation of the stereo head fundamental matrix due to Li et al. 15] However, the former does not seek closed form solutions amenable to error ....

....of the lines. c) Relative error of f 2 (as ) versus fi 1 and fi 2 with Deltau 0 = 10 pixels. Note that f is not computable at fi 1 = Sigmafi 2 . In each case, f has been fixed at 800 pixels in length. stereo head that was previously shown to be a degenerate singularity for selfcalibration [3]. We also observe that accurate self calibration will typically require that vergence angles differ by several degrees. Given f equal to 800 pixels and 20 deg, then the relative error of f 2 is shown in Figure 4(b) to vary approximately linearly when Deltau 0 varies in the range ....

M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela, Direct methods for self-calibration of a moving stereo head, in: B. Buxton and R. Cipolla, editors, Proc. European Conference on Computer Vision (Cambridge, UK, 1996), vol. 2, pp. 415--426.

....matrix. This by itself is not sufficient, however a complete analysis of exactly what configurations will achieve this is not attempted here: we simply note that they are likely to be rather special configurations that are already known to be degenerate, such as when there is no rotation (see [1, 2] for a more detailed analysis of these situations) ....

M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela, "Direct Methods for Self-Calibration of a Moving Stereo Head," Proc. European Conference on Computer Vision, Cambridge, UK. 14-18 Apr 1996. To appear.

....shown by Hartley [5] that seven imaging parameters, describing the relative orientation and focal lengths of a stereo system in general position, may be determined automatically from corresponding image points via a process that has been termed selfcalibration. It was later shown by Brooks et al. [1] that these parameters may not be determined if the images are captured by a stereo head. In this situation, only if the focal lengths are assumed identical may the imaging parameters be recovered. In the special case of an isosceles set up, in which left and right cameras verge inwards an ....

.... 2 Self calibration of a moving camera We now consider the stereo problem induced by a single, moving camera with fixed intrinsic parameters (equivalent to considering a pair of identical cameras) A notation similar, but not identical, to that of Faugeras et al. 2] is adopted; see [1] for a summary of differences. Assume that a camera views a static scene, moving in such a way that its optical axis is confined to a horizontal ground plane, while its vertical axis is at all times orthogonal to this plane. An image pair generated by such a camera will have a translation vector t ....

[Article contains additional citation context not shown here]

M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela, "Direct methods for self-calibration of a moving stereo head," B. Buxton and R. Cipolla, editors, Proc. European Conference on Computer Vision, Cambridge, UK. Lecture Notes in Computer Science 1065, vol. 2, pp. 415-- 426, 1996.

....absence of motion, this commonly adopted configuration is degenerate in that self calibration may no longer be carried out in the sense of Hartley [4] Direct methods of self calibration are then explored in the context of a moving stereo head. Note that an extended version of this work appears in [1]. 2 Stereo head assembly Consider the special case of a stereo head in which a pair of cameras is mounted on a lateral rig. The cameras are free to vary their angles of vergence. The y axes of the two images are parallel, and are orthogonal to the baseline vector, as depicted in Figure 1. The ....

....equations may be generated. Given the need to eliminate , at most 4 of the 5 parameters may be determined, provided the remaining parameter is known. Further discussion of self calibration and the effect of either the tilting or rotation about the optical axis of one camera is to be found in [1]. 4 Self calibration of a horizontally moving stereo head Having seen that a static stereo head, with coplanar optical axes, is a degenerate configuration for self calibration, we now assess the consequences of moving the head. Specifically, we permit: motion of the head such that the optical ....

[Article contains additional citation context not shown here]

Brooks, M. J., Agapito, L., Huynh, D. Q. and Baumela, L. Direct Methods for Self-Calibration of a Moving Stereo Head. Tech. Rep. 1/96, CSSIP, January, 1996.

No context found.

M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela. Direct methods for self-calibration of a moving stereo head. In Proc. 4th European Conference on Computer Vision, LNCS 1065, Cambridge, pages II:415--426, 1996.

No context found.

M. J. Brooks, L. de Agapito, D. Q. Huynh, and L. Baumela. Direct methods for self-calibration of a moving stereo head. In Proc. 4th European Conference on Computer Vision, LNCS 1065, Cambridge, pages II:415--426, 1996.

No context found.

M.J. Brooks, L. de Agapito, D.Q. Huynh, L. Baumela, Direct methods for self-calibration of a moving stereo head. Proceedings of the European Conference on Computer Vision, Cambridge, 1996, pp. 415-426.

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