L. Darsa and B. Costa, "Multiresolution Representation and Reconstruction of Adaptively Sampled Images," in Proceedings of SIBGRAPI, pp. 321--328, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... space) and also on geometry (object space) They are also useful for an adaptive subdivision of image space for progressive re nement [9] Some of them have been recently applied in the image based rendering eld for weighting pixel colour for reconstruction [10] and adaptive sampling strategies [4, 5], and creating a priority schema for sampling in interactive rendering [13] In this paper we will introduce new measures for supersampling re nement criteria, using both colour and geometry. These measures are based on Information Theory concepts. The new measures will be compared with classic ....

....0:6 b max b min b min b max b avg (2) where min, max, and avg represent the minimum, maximum, and average values respectively for r, g, and b colour components. On the other hand, a useful and simple geometric measure for re nement is depth di erence, used recently in image based rendering [4, 5, 10] and interactive rendering [13] Depth di erence is given by p d = 1 d min dmax (3) where d min and dmax represent minimum and maximum distance. In [11] we presented view dependent information theory quality measures for pixel sampling and scene discretization in atland. These measures are ....

Lucia Darsa and Bruno Costa Silva. Multi-resolution representation and reconstruction of adaptively sampled images. In Proceedings of IX Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI '96), pages 321-328, October 1996.

....k D tree and a measure of variance estimates of the nodes to guide the sampling. They demonstrate a piecewise constant reconstruction based upon the k D cells, but suggest a Delaunay triangulation would provide a higher quality interpolant. Subsequent approaches utilize the Delaunay reconstruction [2, 13] and base the adaptive sampling on vertex color discontinuities and triangle size. A priority queue is maintained to determine which triangles to sample next. Pighin et al. [13] present a more sophisticated constrained triangulation that improves the quality with constraint edges inserted at image ....

....to guide sampling. We first describe the approach in the context of a single view, and then address how view motion is supported in this same infrastructure. 2 Image Reconstruction Given a single view, we construct the 2D Delaunay triangulation of the projected samples. In previous approaches [2, 13], the image plane was used as the projection surface. To avoid having to re construct the mesh each frame, the projected points can be utilized to determine the mesh topology in 2D, while still retaining the 3D information at the vertices. The resulting mesh can be viewed from alternate (nearby) ....

[Article contains additional citation context not shown here]

L. Darsa and B. Costa. Multi-resolution representation and reconstruction of adaptively sampled images. In SIBGRAPI, pages 321--328, 1996.

....In [6] several images were blended to get the new one corresponding to the virtual camera. The color of a pixel of each image to be blended was weighted with a quality factor, which included the cosinus of the angle between the normal to the surface and the viewing direction. Darsa et al. [3] use an adaptive sampling strategy based on a priority schema that takes two factors into consideration, the difference in color between adjacent samples and the size of cells. The Gouraud shaded constant color cells are triangles constructed using a Voronoi diagram or Delaunay triangulation, ....

L. Darsa, and B. Costa. Multi-resolution Representation and Reconstruction of Adaptively Sampled Images. In SIBGRAPI'96, pp. 321--328, 1996.

....of the mesh, and view transformations during motion, re using the same representation between adjacent views. Given a single view, other researchers have chosen to build a two dimensional Delaunay triangulation of the projected samples on the image plane as a solution to the reconstruction problem [2, 10, 13]. While this basic approach provides a reasonable barycentric interpolation of the radiance values, the samples must be re projected and the mesh reconstructed each frame to provide a coherent image during viewer motion. Alternate approaches utilize the projected points to determine the mesh ....

....bit is set if the octant is in the negative side of the z=0 plane, the second order bit is set if the octant is on the negative side of the y=0 plane, and the lowest order bit set similarly for x. Given a 3D point p, the octant identifier I is calculated as in the following pseudo code: I = p[2] 0.0 0:4) p[1] 0.0 0:2) p[0] 0.0 0:1) Once the octant is located, the point is converted into integer barycentric coordinates (a; b; c) relative to the triangle q 0 q 1 q 2 forming the quadtree root for that octant. The point v is projected into the plane of the quadtree root ....

[Article contains additional citation context not shown here]

L. Darsa and B. Costa. Multi-resolution representation and reconstruction of adaptively sampled images. In SIBGRAPI, pages 321--328, 1996.

....the object edges, and its sampling density should be inversely proportional to the depth. The implemented algorithm constructs an image space Delaunay triangulation using a Voronoi diagram to adaptively sample the image based on the depth component, similar to previous work by the authors (Darsa and Costa, 1996). Figure 8 shows an image, its depth image and the corresponding triangulation (viewed from the original point of view) in which the farthest objects are sampled more sparsely, and the areas near the edges are sampled finely. Another triangulation created by inverse projecting depth information ....

Darsa, L. and Costa, B. (1996). Multi-resolution representation and reconstruction of adaptively sampled images. In SIBGRAPI '96 Proceedings, pages 321--328.

No context found.

L. Darsa and B. Costa, "Multiresolution Representation and Reconstruction of Adaptively Sampled Images," in Proceedings of SIBGRAPI, pp. 321--328, 1996.

No context found.

L. Darsa and B. Costa, "Multiresolution Representation and Reconstruction of Adaptively Sampled Images," in Proceedings of SIBGRAPI, pp. 321--328, 1996.

No context found.

Lucia Darsa and Bruno Costa. Multi-resolution representation and reconstruction of adaptively sampled images. In Proceedings of IX Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI'96), pages 321--328, October 1996.

No context found.

L. Darsa and B. Costa. Multi-resolution representation and reconstruction of adaptively sampled images. In Computer Graphics Proceedings (SIGGRAPH 96), Annual Conference Series, pages 321--328, 1996.

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