Anandan, P., Computing Dense Displacement Fields with Confidence Measures in Scenes Containing Occlusion, Proceedings DARPA Image Understanding Workshop, pp. 236-246, 1984.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....displacement fields are shown graphically in Figures 2 and 3. Figure 3: The estimated displacementvectors on a frame of the Tennis table sequence 4 Confidence measure In the past work has been done on the reliability of the estimated displacementvector using matching techniques. Anandan [1] uses the curvature of the matching criterion, the sum of squared differences, for measuring the confidence of the correspondence. Patras [8] expresses the confidence measure in terms of the aposteriori probability of the motion vectors, the whole estimation framework being probabilistic. In our ....

....P x (x# y) y (x# y) 9) The smaller of the two eigenvalue of matrix G can be used as the index of the block capacity for estimating the displacement. If the smaller eigenvalue is 0, the measure confidence is taken to be null. This measure is also suggested in [2] We define the range [0,1] as being the interval of possible values for the confidence measure. On the other hand, the confidence measure should only define the relative confidence of the different blocks. Therefore, it is not necessary to have a unique mapping of the eigenvalues to the confidence measures. For this reason ....

P. Anandan. Computing dense displacement fields with confidence measurement of visual motion. In Proc. in SPIE Conference on Intelligent Robots and Computer Vision, pp. 184--194, 1984.

....it requires reliable feature extraction and it fails to describe large areas of the image. Another line of work has addressed the problem of extracting dense displacement or depth estimates from image sequences. However, these previous approaches have either been restricted to twoframe analysis [1] or have used batch processing of the image sequence, for example via spatiotemporal filtering [11] In this paper we introduce a new, pixed based (iconic) approach to incremental depth estimation and compare it mathematically and experimentally to a feature based approach we developed previously ....

....[6, 13, 33] We will specify this in terms of a translational velocity T and an angular velocity R. In the camera coordinate frame (figure 2) the motion of a 3D point P is described by the equation dP T RxP dt Expanding this into components yields dX dt = RyZ R,Y dY dt = R:X RZ [1] dZ dt = T RxY R, X Now, projecting (X, Y, Z) onto an ideal, unit focal length image, X Y x yZ taking the derivatives of (x, y) with respect to time, and substituting in from equation (1) leads to the familiar equations of optical flow [33] Ay Z 0 1 I21 (1 y2) xy x These ....

[Article contains additional citation context not shown here]

P. Anandan, "Computing dense displacement fields with confidence measures in scenes containing occlusion." Proc. DARPA Image Understanding Workshop, pp. 236-246, 1984.

....regularization (non parametric) schemes and allow the estimation of motion in regions with only sparse image structure. Finally, in cases where the flow estimates are poor or uncertain, it is useful to have an estimate of the flow vector s certainty through the use of confidence measures [37], probabilistic estimates [38] or estimation theoretic techniques [39] This work is an important contribution, for it may allow processes that use optic flow to ignore poor measurements and hence produce accurate results. 6.2. Temporal Persistence The second main criticism of optical flow ....

P. Anandan, Computing dense displacement fields with confidence measures in scenes containing occlusion, SP1Elntell. Robots Comput. Vision 521, 1984, 184-194.

....quently, the results of these methods are instantaneous and make no use of the redundancy inherent in a series of frames. Examples of such instantaneous analyses include: estimation of the optical flow (Hildreth [19] 20] Horn and Schunck [27] Nagel and Enkelmann [40] 39] 41] Anandan [2], Fennema and Thompson [12] recovery of motion parameters (Longuet Higgins and Prazdny [35] Negahdaripour, Weldon and Horn [42] 28] Weng, Huang and Ahuja [64] Fennema and Thompson [12] Tsai and Huang [55] 56] Waxman and Wohn [63] Roach and Aggarwal [47] Prazdny [45] Waxman and Ullman ....

....has proven to be effective. Comparisons of typical correlation based marchers can be found in Hannah [17] and Burr, Yen and Xu [9] An in depth study of the applications of these techniques to the estimation of displacement fields in motion sequences is presented by Ariandan [2]. In particular, correlation based estimation has proven to be useful when large interframe displacements occur. On the other hand, the method encounters difficulties with foreshortening. We employ the sum of squared differences (SSD) technique which we briefly describe below and in figure (12) ....

[Article contains additional citation context not shown here]

P. Anandan. Computing Dense Displacement Fields with Confidence Measures in Scenes Containing Occlusion. COINS Technical Report 84-32, University of Massachusetts, Amherst, December 1984.

....quently, the results of these methods are instantaneous and make no use of the redundancy inherent in a series of frames. Examples of such instantaneous analyses include: estimation of the optical flow (Hildreth [19] 20] Horn and Schunck [27] Nagel and Enkelmann [40] 39] 41] Anandan [2], Fennema and Thompson [12] recovery of motion parameters (Longuet Higgins and Prazdny [35] Negahdaripour, Weldon and Horn [42] 28] Weng, Huang and Ahuja [64] Fennema and Thompson [12] Tsai and Huang [55] 56] Waxman and Wohn [63] Roach and Aggarwal [47] Prazdny [45] Waxman and Unman ....

....has proven to be effective. Comparisons of typical correlation based matchefs can be found in Hannah [17] and Butt, Yen and Xu [9] An in depth study of the applications of these techniques to the estimation of displacement fields in motion sequences is presented by Anandan [2]. In particular, correlation based estimation has proven to be useful when large interframe displacements occur. On the other hand, the method encounters difficulties with foreshortening. We employ the sum of squared differences (SSD) technique which we briefly describe below and in figure (12) ....

[Article contains additional citation context not shown here]

P. Anandan. Computing Dense Displacement Fields with Confidence Measures in Scenes Containing Occlusion. COINS Technical Report 84-32, University of Mas- sachusetts, Amherst, December 1984.

....problem [11] Finally, the optimum and inverse Hessian of the resultant quadratic are used as the measurement and its variance, respectively. A similar error analysis was performed by Szeliski [10] for optical flow measurements based on the sum of squared differences (SSD) technique of Anandan [12]. The SSD measure is a simplified version of normalized cross correlation. Our second intensity based similarity measure approximates SSD with the windowing function reduced to a single point 2 ; its associated likelihood is expressed as follows: p(fI T ; I R gjffi) Y x2 Omega R ....

P. Anandan, "Computing dense displacement fields with confidence measures in scenes containing occlusion," in Proc Image Understanding Workshop, Miami Beach, FL:Science Applications International Corporation, pp. 186--196, 1984.

....due to illumination effects, for example, without affecting a phase based velocity estimate. Moreover, a phase based approach have been shown to be more robust when dealing with motions that are not strictly translational [5] Area based methods Area based methods, in the form of block matching [6, 7], is perhaps the most widespread technique for interframe motion estimation, where the assumption is that motion is purely translational and that the motion vector remains unchanged over a particular block of pixels. Motion estimation of this type is, for example, used in most MPEG implementations ....

....of the image intensity function which relates this method to the differential methods described above [3] To minimise the SSD (6) it is necessary to perform some sort of spatial search to determine the motion vector for a given region. Searching procedures in use include finite element approaches [7] and relaxation procedures [6] Also, the majority of block matching implementations, for example most MPEG implementations, assume a fixed block size which imposes a limit on the spatial resolution of the motion estimates obtained [2] Feature based methods In a feature based approach the basic ....

P. Anandan, "Computing Dense Displacement Fields With Confidence Measures in Scenes Containing Occlusion," in Proceedings DARPA Image Understanding Workshop, pp. 236--246, 1984.

.... of the National Conference on Artificial Intelligence, AAAI 90 , Boston, Mass, 1990 Constraints for the Early Detection of Discontinuity from Motion Michael J. Black and P. Anandan Department of Computer Science Yale University New Haven, CT 06520 2158 Abstract Surface discontinuities are detected in a sequence of images by exploiting physical constraints at early ....

....of the National Conference on Artificial Intelligence, AAAI 90 , Boston, Mass, 1990 Constraints for the Early Detection of Discontinuity from Motion Michael J. Black and P. Anandan Department of Computer Science Yale University New Haven, CT 06520 2158 Abstract Surface discontinuities are detected in a sequence of images by exploiting physical constraints at early stages in the processing of visual motion. To achieve accurate early discontinuity detection we exploit five physical constraints on the presence of discontinuities: the ....

[Article contains additional citation context not shown here]

P. Anandan. Computing dense displacement fields with confidence measures in scenes containing occlusion. In SPIE fnt. Conf. Robots and Computer Vision, 51, pages 184-194, 1984.

No context found.

Anandan, P., Computing Dense Displacement Fields with Confidence Measures in Scenes Containing Occlusion, Proceedings DARPA Image Understanding Workshop, pp. 236-246, 1984.

No context found.

P. Anandan. Computing Dense Displacement Fields with Confidence Measures in Scenes Containing Occlusion. In SPIE Intelligent Robots and Computer Vision, volume 521, pages 184--194, 1984.

No context found.

P. Anandan. Computing dense displacement fields with confidence measures in scenes containing occlusion. In Proceedings of the SPIE: Intelligent Robots and Computer Vision, volume 521, pages 184--194. Cambridge, MA, 5--8 Nov. 1984.

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