Quam, Hierarchical warp stereo. Tech. Rep. Center, International, 1986.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... whose behavior mimics closely that of iterative optimization algorithms [73, 97, 132] Hierarchical (coarse to fine) algorithms resemble such iterative algorithms, but typically operate on an image pyramid, where results from coarser levels are used to constrain a more local search at finer levels [126, 90, 11]. The most common pixel based matching costs include squared intensity differences (SD) 51, 1, 77, 107] and absolute intensity differences (AD) 58] In the video processing community, these matching criteria are referred to as the mean squared error (MSE) and mean absolute difference (MAD) ....

....methods inspired by classic (infinitesimal) optic flow computation. Here, images are successively warped and motion estimates incrementally updated until a satisfactory registration is achieved. These techniques are most often implemented within a coarse to fine hierarchical refinement framework [90, 11, 8, 112]. A univalued representation of the disparity map is also not essential. Multi valued representations, which can represent several depth values along each line of sight, have been extensively studied recently, especially for large multiview data set. Many of these techniques use a voxel based ....

[Article contains additional citation context not shown here]

L. H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149--155, New Orleans, Louisiana, 1984. Science Applications International Corporation.

.... algorithms that do not explicitly state a global function that is to be minimized, but whose behavior mimics closely that of iterative optimization algorithms [46, 63, 83] Hierarchical (coarse to fine) algorithms resemble such iterative algorithms, but typically operate on an image pyramid [80, 58, 7]. 3.1. Matching cost computation The most common pixel based matching costs include squared intensity differences (SSD) 34, 1, 48, 68] and absolute intensity differences (SAD) 39] More recently, robust measures, including truncated quadratics and contaminated Gaussians have been proposed ....

....methods inspired by classic (infinitesimal) optic flow computation. Here, images are successively warped and motion estimates incrementally updated until a satisfactory registration is achieved. These techniques are most often implemented within a coarse to fine hierarchical re finement framework [58, 7, 5, 70]. A univalued representation of the disparity map is also not essential. Multi valued representations, which can represent several depth values along each line of sight, have been extensively studied recently, especially for large multiview data set. Many of these techniques use a voxel based ....

L.H. Quam. Hierarchical warp stereo. In IUW, pp. 149-155, 1984.

....two very different viewpoints [6, 20, 24, 25] but they did not consider a large change in resolution. The use of scale space in conjunction with stereo matching has been restricted to hierarchical matching: correspondences obtained at low resolution constrain the search space at higher resolutions [7, 21, 14]. Scale space properties are thoroughly studied in [15] and the same author attempted to characterize the best scale at which an image feature should be represented [16] A similar idea is presented in [17] to detect stable points in scale space. Our work is closely related to [9] who attempts to ....

L. H. Quam. Hierarchical warp stereo. In Reading in computer Vision, pages 80-86. Morgan Kaufman, 1987.

....themselves [14,15] Jaillon and Montavert [16] show how to register when knowledge of the three dimensional structure of the scene is available. Since a true 3D solution is difficult, in many cases a 2D registration between pixels is sought. Finding this may be achieved by optical flow methods [18,23], but this does not produce registrations accurate enough for video mosaicing. Aiger and Cohen [ 19] used an iterative algorithm to improve the registration accuracy of the optical flow solution. Such solutions are common in three dimensional imaging for medical applications, where the quality of ....

....transformation consisting of translation only as this initial guess (only rn 2 and rn 5 are non zero) implicitly assuming that the translation parameters in the transformation matrix are significantly larger than the others. This translation may be found efficiently using the multi pyramid method [23]. 3. Registering an Image Sequence This section elaborates on how the basic two image registration method is extended to handle a long image sequence. 3.1. The Sequential Algorithm In the sequential extension of the basic two image registration procedure, each image is registered to its ....

L.H. Quam. "Hierarchical warp stereo." In Image Understanding Workshop, pp. 149155, December 1984. 13

....to find the best registration. We have implemented two different techniques for dealing with this problem. The first technique, which is commonly used in computer vision, is hierarchical matching, which first registers smaller, subsampled versions of the images where the apparent motion is smaller [Quam, 1984; Witkin et al. 1987; Bergen et al. 1992] Motion estimates from these smaller coarser levels are then used to initialize motion estimates at finer levels, thereby avoiding the local minimum problem (see [Szeliski and Coughlan, 1994] for details. While this technique is not guaranteed to find ....

L. H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149--155, Science Applications International Corporation, New Orleans, Louisiana, December 1984.

....especially when the initial misregistration is more than a few pixels. A useful heuristic for enlarging the region of convergence is to use a hierarchical or coarse to fine algorithm, where estimates from coarser levels of the pyramid are used to initialize the registration at finer levels [Qua84, Ana89, BAHH92] This is a remarkably e#ective technique, and we typically always use 3 or 4 pyramid levels in our mosaic construction algorithm. However, it may still sometimes fail if the amount of misregistration exceeds the scale at which significant image details exist (i.e. because these ....

L. H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149--155, New Orleans, Louisiana, December 1984. Science Applications International Corporation.

....been shown that a non linear criterion minimization using the LevenbergMarquardt method yields very good results [Sze94] but it is very sensitive to the local minima and computationally expensive. In another case when the overlap is smaller we can use a hierarchical matching to avoid local minima[Qua84, WTK87, BAHH92] For larger camera motion the phase correlation method has been used [KH75, Bro92] However for large rotations around the optical axis, very few methods are eOEcient. Among the best methods we nd the work of Dani and Chaudhuri: Their method works for up to 15 degrees rotations ....

L.H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149155, New Orleans, Louisiana, december 1984. SAIC.

....will typically give erroneous results in these situations. For optical flow algorithms based on matching, the errors appear as false matches . A number of authors have developed coarse to fine processing strategies for handling the temporal aliasing problem in the context of motion estimation [7, 9, 5, 1]. These algorithms are typically efficiently implemented using image pyramids [3] The motivating concept in these approaches is that the aliasing only affects the high spatial frequencies of the input sequence. Thus, one can accurately estimate image velocities of a spatially lowpass filtered ....

L. Quam. Hierarchical warp stereo. In Proceedings of the DARPA Image Understanding Workshop, September 1984.

....can result in a greater number of false positives in occlusion zones and increased smoothing of disparity across discontinuities, although the number of false negatives due to noise and outliers may decrease. Different approaches have been developed to tackle individual issues (for example, 9] [10]) within the framework of linear correlation measures. For instance, Quam [10] addresses the 0162 8828 98 10.00 1998 IEEE ################ .# D.N. Bhat is with LG Electronics Research Center of America, 40 Washington Road, Princeton, NJ 08550. E mail: dbhat lgerca.com. # S.K. Nayar is with ....

....smoothing of disparity across discontinuities, although the number of false negatives due to noise and outliers may decrease. Different approaches have been developed to tackle individual issues (for example, 9] 10] within the framework of linear correlation measures. For instance, Quam [10] addresses the 0162 8828 98 10.00 1998 IEEE ################ .# D.N. Bhat is with LG Electronics Research Center of America, 40 Washington Road, Princeton, NJ 08550. E mail: dbhat lgerca.com. # S.K. Nayar is with the Department of Computer Science, Columbia University, New York, NY 10027. ....

[Article contains additional citation context not shown here]

# L.H. Quam, "Hierarchical Warp Stereo," Proc. ARPA Image Understanding Workshop, pp. 149-155, 1984.

...., I s x y z , 2) I a D x y z , I a D x y z , T dD (3) 30 Chapter 2 3 D Hierarchical Deformable Registration is the first order derivative, i.e. image gradient, at voxel in the atlas. Substitute (3) into (2) gives the image brightness constraint [77]: Figure 18: Intensity histograms of corpus callosum (top) in the atlas (dotted line) and the subject s volume (solid line) as well as the corresponding ones of skull (bottom) 0 50 100 150 200 250 300 0 0.01 0.02 0.03 0.04 Corpus Collasum Intensity Histogram Intensity atlas patient ....

Quam, "Hierarchical warp stereo", Image Understanding Workshop, pp 149-155, December, 1984.

....to provide a good first guess of the matrix coefficients for the algorithms to work. The first order approximation is obtained by solving the simpler problem of translation between the two pictures. This can be done in one of two ways: 1) Phase corelation [14] 2) Pyramid Hierarchical Matching [15]. The results are then used as the initial coefficients of the projective matrix. Stitching pictures together. Now that we have the projective matrix between each two adjacent pictures of a planar scene, we can proceed and stitch the scene together into one picture. For each picture (exept the ....

L.H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149-155, Science Applications International Corporation, New Orleans, Louisiana, December 1984.

.... gradient technique will fail if there is too much spatio temporal aliasing (i.e. if the displacements being measured are greater than one half of a cycle of the highest spatial frequencies present in the pre filtered image sequence) Similar multiscale warping approaches have been used by Quam [7] and Anandan [3] We first built a (spatial) gaussian pyramid [8] on each frame of the image sequence: Each frame was recursively blurred using a simple gaussian filter and subsampled by a factor of two in each spatial direction. This resulted in a set of images (a pyramid ) of different ....

Lyn Quam. Hierarchical warp stereo. In Proceedings of the DARPA Image Understanding Workshop, September 1984.

.... technique will fail if there is too much spatio temporal aliasing (i.e. if the displacements being measured are greater than one half of a cycle of the highest spatial frequencies present in the pre filtered image sequence) Similar multi scale warping approaches have been used by Quam [16] and Anandan [12] We first built a (spatial) gaussian pyramid [17] on each frame of the image sequence: Each frame was recursively blurred using a simple gaussian filter and subsampled by a factor of two in each spatial direction. This resulted in a set of images (a pyramid ) of different ....

Lyn Quam. Hierarchical warp stereo. In Proceedings of the DARPA Image Understanding Workshop, September 1984.

....between a point s location along the scanline in the resampled reference image and in the resampled target images. From the disparity map and orientation parameters, a DEM can easily be computed using simple geometric relations (see Figure 8) Terrest employs a hierarchical search technique (Quam, 1984, Schultz, 1995) which is widely used to reconstruct the shape of natural terrain. Such techniques avoid searching the entire scan line in the target image by first looking for the disparities associated with large objects at the low resolution levels of the hierarchy, and then gradually refining ....

Quam, L. H., 1984. Hierarchical Warp Stereo. In: Proceedings: Image Understanding Workshop, pp. 149--155.

....onto the facets of the surface. Consequently, the reconstruction can be significantly more accurate for slanted surfaces. Some correlation based algorithms achieve similar results by using variable shaped windows in the images. Control Data s work (Panton 1978) the Hierarchical Warp Stereo System (Quam 1984), Nishihara s real time stereo matcher (1984) and the adaptative windows technique described in (Kanade and Okutomi 1990) are examples of such methods. However, they typically use only image centered representations of the surface. As for the monocular information source, we have chosen to use ....

Quam, L. (1984). Hierarchical warp stereo. In ARPA Image Understanding Workshop, pages 149--155.

....others [Hanna, 1991; Bergen et al. 1992] In this paper, we use projective descriptions of motion and depth [Faugeras, 1992; Mohr et al. 1993; Szeliski and Kang, 1994] for our constrained motion model, which removes the need for calibrated cameras. Stereo matching [Barnard and Fischler, 1982; Quam, 1984; Dhond and Aggarwal, 1989] is traditionally considered as a separate sub discipline within computer vision (and, of course, photogrammetry) but there are strong connections between the two problems. Stereo can be viewed as a simplified version of the constrained motion model where the egomotion ....

.... of images, more recent algorithms use sequences of images (multiframe stereo or motion stereo) Bolles et al. 1987; Matthies et al. 1989; Okutomi and Kanade, 1992; Okutomi and Kanade, 1993] Hierarchical (coarse to fine) matching algorithms have a long history of use both in stereo matching [Quam, 1984; Witkin et al. 1987] and in motion estimation [Enkelmann, 1988; Anandan, 1989; Singh, 1990; Bergen et al. 1992] Hierarchical algorithms first solve the matching problem on smaller, lower resolution images and then use these to initialize higher resolution estimates. Their advantages include ....

L. H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149--155, Science Applications International Corporation, New Orleans, Louisiana, December 1984.

....motion patches based on estimates of the current residual error in the flow estimate [Muller et al. 1994] Our approach is similar to this latter work, except that it preserves inter patch motion continuity, and uses both split and merge techniques. Stereo matching [Barnard and Fischler, 1982; Quam, 1984; Dhond and Aggarwal, 1989] is traditionally considered as a separate sub discipline within computer vision (and, of course, photogrammetry) but there are strong connections between it and motion estimation. Stereo can be viewed as a simplified version of constrained motion estimation where the ....

....window sizes in stereo [Okutomi and Kanade, 1992; Okutomi and Kanade, 1994] is similar in spirit to the idea used in this paper, although their algorithm has a much higher computational complexity. Hierarchical (coarse to fine) matching algorithms have a long history of use both in stereo matching [Quam, 1984; Witkin et al. 1987] and in motion estimation [Enkelmann, 1988; Anandan, 1989; Singh, 1990; Bergen et al. 1992] Hierarchical algorithms first solve the matching problem on 4 3 General problem formulation smaller, lower resolution images and then use these to initialize higher resolution ....

[Article contains additional citation context not shown here]

L. H. Quam. Hierarchical warp stereo. In Image Understanding Workshop, pages 149--155, Science Applications International Corporation, New Orleans, Louisiana, December 1984.

....some psychophysical results of stereo vision (Lehky and Sejnowski, 1990) It remains to be demonstrated if these cells can be used to compute disparity maps from stereograms. Many models for disparity computation have been proposed over the years (Marr and Poggio, 1976; Marr and Poggio, 1979; Quam, 1984; Prazdny, 1985; Pollard, Mayhew and Frisby, 1985; Qian and Sejnowski, 1988; Sanger, 1988; Yeshurun and Schwartz, 1989) Unfortunately, most of them are non biological, either because they require sharply disparitytuned units with preferred disparities covering a wide range of values, or because ....

Quam, L. H. (1984). Hierarchical warp stereo, Proceedings of the DARPA Image Understanding Workshop, pp. 149--155.

....to the image plane. Instead, we compare the intensities as projected onto the facets of the surface. Consequently, the reconstruction can be significantly more accurate for slanted surfaces Some correlation based algorithms achieve similar results by using variable shaped windows in the images [15, 14, 11, 1, 4]. However, they typically use only image centered representations of the surface. As for the monocular information source, we have chosen to use shading, where shading is the change in image intensity due to the orientation of the surface relative to a light source. The main reason for this is ....

....even if taken from widely differing viewpoints. It accommodates such viewpoint dependent effects as selfocclusion and self shadowing. ffl Our technique for doing stereo avoids the constant depth assumption of traditional correlation based stereo algorithms. As the Hierarchical Warp Stereo System [15], it effectively uses variable sized windows in the images but does it in a more generic fashion. ffl Our approach to shape from shading is applicable to surfaces with slowly varying albedo. This is a significant advance over traditional approaches that require constant albedo. ffl We have ....

L.H. Quam. Hierarchical warp stereo. In ARPA Image Understanding Workshop, pages 149--155, 1984.

No context found.

Quam, Hierarchical warp stereo. Tech. Rep. Center, International, 1986.

No context found.

L. H. Quam. Hierarchical warp stereo. Technical Report 402, AI Center, SRI International, 1986.

No context found.

L.H. Quam, Hierarchical warp stereo, in: Reading in computer Vision, Morgan Kaufman, 1987, pp. 80--86.

No context found.

Quam LH (1984) Hierarchical Warp Stereo. In: Proc. DARPA Image Understanding Workshop, New Orleans, La. Defence Advanced Research Projects Agency, USA, pp 149--155

No context found.

Lyn Quam. Hierarchical warp stereo. In Proceedings of the DARPA Image Understanding Workshop, pages 149-155, September 1984.

No context found.

Learning. Quam, L. H. (1984). "Hierarchical Warp Stereo," Proc. DARPA IUW, New Orleans, pp. 149-155.

First 50 documents

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