A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....or thin plate splines [11] Orientation correlation finds translational transformations. This is a prominent transformation in many of the above applications. Applications which require a more complex transformation may still use a translational model as the first stage of the estimation process [1, 8]. For evaluation of a transformation two matching methodologies are prevalent in the literature; area based methods (also known as direct methods) and feature based methods. Area based methods match measurable image quantities, e.g. brightness [16] or phase [17] 12] Feature based methods match ....

....in the literature; area based methods (also known as direct methods) and feature based methods. Area based methods match measurable image quantities, e.g. brightness [16] or phase [17] 12] Feature based methods match features extracted from the images, e.g. corners [14] lines [9] or junctions [8]. Orientation correlation matches the feature of gradient orientation for each pixel. Orientation correlation is a feature based method with many of the advantageous properties of area based methods. With the transformation selected and method of evaluating a transformation defined, image ....

A. Can, C.V. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proc. IEEE Conf. on Computer Vision and Pattern Recognition, pages 286--292, June 1999.

....must be identified on all of the images. This will generally require warping the images to a common standard orthorectification. Although techniques for automatic image registration are known [FM96, SD96, DKC96] it is nontrivial to achieve pixel level matching (although this is improving, see [CSR99] However, as discussed in the preceding paragraph and shown in the example in Section 5.2, pixel level resolution is not usually necessary. Therefore registration is only needed to within a few pixels, based on the size of a location or feature. The ability to work with poorer than ....

A. Can, C.V. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, June 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proc. CVPR, pages 286--292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 286--292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

No context found.

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

....by using automated, image based tools that determine the locations of selected spots on the retina surface using clinical image sequences. We call this capability spatial referencing. The reference coordinate frame is a spatial map of the retina, pre computed from o line diagnostic images [8, 9, 10, 11] (Figure 1) On line spatial referencing must be fast enough to enable real time control decisions that are the key to developing diagnostic and treatment aids. An important application for spatial referencing is laser retinal surgery. This is the only proven treatment for leading ....

....images and associated digital distance maps, and inter image transformations between diagnostic images. 2 Transformation Models Three transformation models are used in the indexing, re nement and veri cation procedures. The nal and most accurate model is a quadratic model derived in [8, 11]. Let x = x; y) be the coordinates of a point location in one image, I p , and let u = u; v) be the coordinates of the same location on the retina in a second image, I q . Then, the Figure 3: Illustrating indexing. A vector, f , of quasi invariants is computed from a constellation of ....

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proceedings IEEE Conference on Computer Vision and Pattern Recognition, pages 286-292, 1999.

....of these points, if known to be stable, are valuable as features (i.e. landmarks) for image registration and mosaicing. In retinal imaging, these points are known to be stable unless the retina is detached. The often unique pattern of angles of intersection can be used as landmark signatures [20, 28, 29, 30, 31, 32]. Registered images can be used to reveal retinal changes and pathologies [33, 28] Mosaics of retinal images provide high resolution, wide extent imaging of the retina for diagnosis of pathologies of the retinal periphery [29, 30, 34] High speed image registration can provide the basis for ....

.... 10241024 pixels, and were captured using a Topcon IMAGENET digital camera system at the Center for Sight (Albany, New York) The method for measuring the accuracy and repeatability of landmark estimation is based on a highly accurate image registration algorithm developed in prior work [29, 30, 31, 32] is used as a testbed and standard for measuring landmark estimation errors. This is a hierarchical registration technique that uses correspondences between landmarks in two different retinal images to estimate a 12 parameter, quadratic transformation mapping one retinal image onto another ....

[Article contains additional citation context not shown here]

Can, A., C. V. Stewart, B. Roysam, "Robust Hierarchical Algorithm for Constructing a Mosaic from Images of the Curved Human Retina," Proceedings IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Fort Collins, Colorado, June 1999.

....frame, and Fig. le shows the vascular landmarks (crossovers and bifurcations) for this sample frame. These landmarks are matched for image registration in the context of spatial mapping and real time spatial reckoning [6,3,7] relative to a previously constructed mosaic map of the entire retina [8]. This is an instance of a hard real time system [9] in which the computations must be completed prior to a deadline, else the system is considered to have failed. A failure represents a loss of tracking, requiring the surgical laser to be switched off. System performance degrades with an ....

Can, A., Stewart, C. V., and Roysam, B., "Robust Hierarchical Algorithm for Constructing a Mosaic from Images of the Curved Human Retina," Proc. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Vol. 1I, 286-292, Fort Collins, Colorado, June 1999.

....safety shutoffs when the laser is aimed incorrectly. This can dramatically improve the treatment of many of the leading causes of blindness affecting tens of millions of patients [21] Constructing an accurate mosaic is therefore an important milestone in this project. In recently published work [6], we described a retinal mosaic construction method based on pairwise registration of retinal images with a central anchor image. The registration algorithm uses a 12 parameter image to image transformation model that accounts for the unknown curvature of the retinal surface. The algorithm is ....

....while still providing enough constraints for joint estimation. The overall results are illustrated in Figure 5 using anchor frames on the periphery of the retina. 3 Pairwise Registration This section summarizes our current technique for computing image to image (pairwise) transformations [6]. 3.1 Transformation Model In the image to image transformation model, the retina is modeled as a quadratic surface. Rigid transformations between viewpoints are assumed, and the cameras are weakperspective. Let I n be one image frame, let I r be the reference image frame, and let p = x;y) T ....

[Article contains additional citation context not shown here]

A. Can, C. Stewart, and B. Roysam. Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina. In Proc. CVPR, pages 286--292, 1999.

No context found.

A. Can, C. V. Stewart, and B. Roysam, "Robust hierarchical algorithm for constructing a mosaic from images of the curved human retina," in Proc. IEEE Conf. Comput. Vision Pattern Recogn., 1999, pp. 286--292.

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