Donald B. Gennery. Visual tracking of known three--dimensional objects. International Journal of Computer Vision, 7(3):243--270, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....two main classes: feature based and modelbased. The former approach tracks features such as geometrical primitives (points, segments, circles, object contours [1, 10] regions of interest [8] The latter explicitly uses a model of the tracked objects. This model can be a CAD model [4, 6, 13, 16, 14] or a 2D template of the object [12] This second class of methods usually provides with a more robust solution (for example, it can cope with partial occlusion of the objects) Both approaches may use Kalman filters to predict and estimate the position of the tracked primitives over time. In our ....

D. Gennery. Visual tracking of known three-dimensional objects. Int. J. of Computer Vision, 7(3):243--270, 1992.

....that of the scene is unimportant. It is well known that relatively few correspondences between the image and the scene are necessary to constrain the relative pose of a camera and a known object or scene [22] Fairly general 3D solutions for finding the relative pose have been known for some time[16]. 1.3.6 Prior Image Based Rendering Work Since we will render predicted imagery, this work is related to image based rendering. The problem of image based rendering [9] 27] 49] is related to visual tracking in that the motions and deformations of all regions of the image are being predicted. ....

D. B. Gennery. Visual tracking of known three-dimensional objects. Int'l Journal of Computer Vision, 7(3):243-270, 1992.

.... robots and not to the human operator Two important choices in the design of the visual pose recovery system were, first, whether to use data driven structural recovery methods for polyhedra such as used in the system developed by Barth et al. 13] or to rely on model based, or prior based methods [17, 18, 19, 20, 21, 22, 23, 24]; and second, whether to use calibrated vision to recover Euclidean structure or un or partially calibrated vision to recover structure modulo an unknown transformation. For the latter choice, whilst there are certainly tasks in hand eye coordination for which uncalibrated vision is quite ....

....to multiple cameras and performance improvements by the use of robust methods. Two important choices in the design of the vision system for the work conducted in this thesis were, first, whether to use data driven structural recovery methods for polyhedra (e.g. 106,107] or model based methods [17,18,19,20,21,22,23,24] and, secondly, whether to use calibrated vision to recover Euclidean structure or un or partially calibrated vision to recover structure modulo an unknown transformation. For the latter choice, whilst there are certainly tasks in hand eye coordination for which uncalibrated vision is quite ....

[Article contains additional citation context not shown here]

D.B. Gennery. Visual tracking of known three-dimensional objects. International Journal of Computer Vision, 7(3):243--270, 1992.

....local search allows the tracker to run quickly, and overcomes any problems establishing correspondence between model and image features. The pose estimation problem is typically over constrained, in that there are usually more measurements available than are required. Recent tracking algorithms [6, 7, 11] have responded to this redundancy by employing all the measurements and some sort of leastsquares pose fitting criterion. The problem with such approaches is that they are easily deceived by the sort of rogue measurements so often produced by visual feature detectors. Moreover, a poor pose ....

....problem could be tackled using some sort of least squares technique, perhaps weighting each point s influence by the amount of confidence we have in that particular measurement. Indeed, a least squares approach is often taken in object recognition and tracking systems: see, for example, [6, 7, 11]. The problem with least squares techniques is that they are easily deceived by a rogue measurement. Suppose that one of the point detectors produced a gross error (which is bound to happen eventually with simple feature detectors) as in Figure l(a) The least squares solution for the object s ....

D. Gennery. Visual tracking of known three-dimensional objects. International Journal of Computer Vision, 7(3):243 270, 1992.

....etc. Automatic road following has been accomplished by tracking the edges of the road [34] Various snake like trackers are used to track objects in 2D as they move across the camera image [2, 8, 11, 29, 46, 49, 54] Three dimensional models, while more complex, allow for precise pose estimation [17, 31]. The key problem in model based tracking is to integrate simple features into a consistent whole, both to predict the configuration of features in the future and to evaluate the accuracy of any single feature. While the list of tracking applications is long, the features used in these ....

....consistent whole, both to predict the configuration of features in the future and to evaluate the accuracy of any single feature. While the list of tracking applications is long, the features used in these applications are vari ations on a very small set of primitives: edgeis or line segments [12, 17, 30, 31, 43, 45, 49, 57], corners based on line segments [23, 41] small patches of texture [13] and easily detectable highlights [4, 39] Although the basic tracking principles for such simple features have been known for some time, experience has shown that tracking them is most effective when strong geometric, ....

D. B. Gennery. Visual tracking of known three-dimensional objects. Itt. J. Computer Visiot, 7(3):243 270, 1992.

....about the frame to frame motion of the target. Many works, including the one described in this paper, assume spatio temporal continuity; i.e. the location of the target in incoming images can be approximately predicted using the previous location and the estimated motion of the target. References [2] and [3] 4] are also examples of work that make this assumption. In another method, 5] described as robust to large and arbitrary motions between images, line detection is performed on the raw image. The extracted lines are then compared with the model, and a best first search algorithm is used ....

D. B. Gennery. "Visual Tracking of Known Three-Dimensional Objects." International Journal of Computer Vision. 7:3, pp. 243 - 270. 1992.

....Video Board. A PUMA 560, 6 DOF industrial robot, is used for the haptic device. We put an aluminum plate with four markers, small incandescent lamps covered by translucent lenses, at the tip of the PUMA for tracking. We implemented the pose estimation algorithm based on the extended Kalman filter[6]. We first took the relative pose between the camera and the plate as the state variables of the estimator. The motion of the haptic device, however, affects the state estimator and the background image tends to be shaky even when the camera display system is stationary. Since we know exactly the ....

D. B. Gennery, "Visual Tracking of Known ThreeDimensional Objects", Int. J. of Computer Vision, vol.7, no.3, pp.243 270 (1992)

....such as human computer interaction and visual surveillance. An aligned image according to the recovered head motion would facilitate facial expression analysis and face recognition. Many approaches have been proposed to recover 3D head motion. One approach is to use distinct image features [2,4,7,18], which works well when the features are reliably tracked over the image sequence. When good feature correspondences over the entire sequence are not available, tracking the entire head region using a 3D head model is more reliable. One can use an anatomically based model for head motion recovery. ....

D.B. Gennery. Visual tracking of known three-dimensional objects. In IJCV, vol. 7, no. 3, pp. 243-270, 1992.

....of a solid object, or a kinematic and shape model of a structured object (e.g. the human body) we can manipulate it to fit the motion of the model to an object in the video using any prediction method (e.g. a Kalman filter) The model and the scene are usually compared using edge features. [8] deals with the problem of tracking objects with known 3D shapes. Shape constraints provide more information about the object configuration or the imaging conditions than point features; the deformation of shapes under changes in object pose or camera parameters (e.g. focal length) provides ....

D.B. Gennery, "Visual Tracking of Known ThreeDimensional Objects," IJCV, Vol. 7, pp. 243-270, 1992.

No context found.

Donald B. Gennery. Visual tracking of known three--dimensional objects. International Journal of Computer Vision, 7(3):243--270, 1992.

No context found.

D.B. Gennery, "Visual tracking of known three dimensional objects," International Journal of Computer Vision, vol. 7, no. 3, pp. 243--270, 1992.

No context found.

D. B. Gennery, Visual tracking of known three-dimensional objects, International Journal of Computer Vision 7 (3) (1992) 243--270.

No context found.

D. Gennery. Visual Tracking of Known Three Dimensional Objects. Int. J. Computer Vision, 7(3), pp 2430270, 1992.

No context found.

D.B. Gennery. Visual tracking of known three-dimensional objects. International Journal of Computer Vision, 7(3):243--270, April 1992.

No context found.

D.B. Gennery. Visual tracking of known three-dimensional objects. Int. J. of Computer Vision, 7(3):243--270, 1992.

No context found.

D.B. Gennery, "Visual tracking of known three dimensional objects," International Journal of Computer Vision, vol. 7, no. 3, pp. 243--270, 1992.

No context found.

D.B. Gennery. Visual tracking of known three-dimensional objects. Int. J. of Computer Vision, 7(3):243--270, 1992.

No context found.

D. Gennery. Visual tracking of known three-dimensional objects. Int. J. of Computer Vision, 7(3):243--270, 1992.

No context found.

D. B. Gennery. Visual Tracking of Known ThreeDimensional Objects. International Journal of Computer Vision, 7(3):243--270, April 1992.

No context found.

D. Gennery, "Visual tracking of known three-dimensional objects," International Journal of Computer Vision, vol. 7, pp. 243--270, 1992.

No context found.

# D.B. Gennery, "Visual Tracking of Known Three-Dimensional Objects," Int'l J. Computer Vision, vol. 7, no. 3, pp. 243--270, 1992.

No context found.

D. B. Gennery, Visual tracking of known three-dimensional objects, International Journal of Computer Vision 7 (3) (1992) 243--270.

No context found.

D. B. Gennery. "Visual tracking of known three-dimensional objects". Int. Journal of Computer Vision, 7(3):243--270, 1992.

No context found.

D. B. Gennery, "Visual tracking of known three-dimensional objects," Int. J. Computer Vision, vol. 7, no. 3, pp. 243--270, 1992.

No context found.

D.B.Gennery. Visual tracking of known three-dimensional objects. Intl. Journal of Comput. Vision, 8:243--270, 1992.

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