ITOH, T., AND KOYAMADA, K. 1995. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics 1, 4 (Dec.), 319--327.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....that intersect the current isosurface. Construction of the isosurface begins at these seeds and propagates through neighboring cells using adjacency and intersection information. The difficult portion of the surface propagation algorithm lies in locating and selecting the seed cells. Itoh et al. [6, 7] find the local extremum points in the data and connect them with a graph in the spatial domain. The seed set consists of the cells containing extremum points, plus all cells intersected by the arcs of the graph and some cells along the boundaries if the volume has holes . A thinning algorithm, ....

Takayuki Itoh and Koji Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....cells can be expensive. To improve the performance, researchers have proposed various schemes that can accelerate the search process. Examples include Wilhelm and Van Gelder s Octrees[2] Livnat et al. s NOISE method[3] Shen et al. s ISSUE algorithm[4] Itoh and Koyamada s Extrema Graph method [5, 6], Bajaj et al. s Fast Isocontouring method [7, 8] and Cignoni et al. s Interval Tree[9] algorithm. Inevitably, these acceleration algorithms incur overhead for storing extra search indices. For a steady scalar field, i.e. only a single time step of data is present, this extra space is often ....

....I O optimal techniques to build the interval tree on disk, and the access of the interval tree is driven by demand. Chiang, Silva, and Schroeder also expanded the I O optimal techniques for out of core isosurface extraction [14] In addition to the space and value partition methods, Itoh et al. [5, 6] and Bajaj et al. 7, 8] proposed algorithms using a surface propagation scheme. In their methods, a small set of seed cells is first extracted; and isosurfaces of any given isovalue can then be computed by propagating surfaces from certain seeds through adjacencies. Bajaj et al. s algorithm is ....

[Article contains additional citation context not shown here]

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4), Dec. 1995.

....Timing Results for the 29g dataset . 103 viii Introduction 1.1 Overview Applications in several disciplines require sampling of some physical quantity in three dimensions, followed by visualization of the data thus acquired. These disciplines include medical imaging [15, 18, 17], fluid dynamics [17] and X ray crystallography [13, 6] The sampling process produces a finite number of data points with associated values. Most, if not all, visualization techniques assume an underlying real valued function, and approximate it by means of some interpolation function f (x) ....

....generated for rendering. Since all algorithms considered generate at most 4 triangles per cell (see Fig. 3.1, p.15 and Fig. 3.5, p.22) these two measures are within a constant factor of each other. Although k = #(N ) in extreme cases, it is claimed that, for typical data , k = O(N ) [16, 15]. Definition 2.10 t is the number of local extrema in the mesh M . A local extremum in the mesh M is a vertex x i , all of whose neighbours either have smaller values than h i (a local minimum Def. 4.9, p.33) or larger values (a local maximum Def. 4.9, p.33) I will show in Sec. 4.5, p.35 ....

[Article contains additional citation context not shown here]

Takayuki Itoh and Koji Koyamada. Automatic Isosurface Propagation Using an Extrema Graph and Sorted Boundary Cell Lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, 1995.

....of values ranging from # min to # max spaced by a ## interval. On a grid made of N c cells and rendered with N # slices, a naive algorithm trying to check each cell against every plane would have an O(N # N c ) complexity, which is not acceptable in an interactive environment. As underlined in [48], this complexity can be reduced to O(N # ) if just the intersected cells are processed for each slice. To this respect, the contour seeds method [39] is probably optimal: for each connected component of the isosurface, a single seed cell and mesh connectivities allow all the intersected ....

....edges are sorted) and duplicates combinatorial information which are often available in CFD grids. Moreover, the structure has to be recomputed if the number of slices changes. Connections between the cells of the mesh can be used more efficiently to traverse the isosurface, as shown in [49] [48], 39] Yet, using connections may not be sufficient to get the correct isosurfaces if the mesh has holes or concavities (Figure 2) Thus, the lazy sweep method [17] updates an active edges list using the list of edges around each vertex. Concavities are handled by checking all the border ....

[Article contains additional citation context not shown here]

T. Itoh and K. Koyamada, "Automatic isosurface propagation using an extrema graph and sorted boundary cell lists," IEEE Trans. Visual. Comput. Graphics, vol. 1, no. 4, pp. 319--327, December 1995.

....seed based and range based. Seed based algorithms make use of a list of isovalues and a set of starting voxels. From these starting points the isosurface is constructed via propagation to adjacent cells. Seed based techniques typically have a preprocess of , where n is the number of voxels [16, 1, 8]. Range based methods can be considered part of the space partition group of algorithms. The first reported technique was based on octrees [18] Later techniques more suited to unstructured grids where developed [5] These techniques group the voxel data into buckets. A similar approach [6] ....

T. Itoh and K. Koyamada. Automatic Isosurface Propagation Using an Extrema Graph and Sorted Boundary Cell Lists. 1(4):319--327, 1995.

....all cells in the volume. An improvement is gained by using octrees [21] to skip cells not contributing to isosurface. Near optimal and optimal algorithms [8, 14, 3, 5] mostly use search structures on the function value domain. Contour propagation is used for efficient isosurface extraction [3, 7, 6]. Given an initial cell that contains the isosurface, the whole surface can be traced by contour propagation. This property is utilized (e.g. in the seed sets) to significantly reduce the space and time for searching isosurface cells. In addition, it allows identifying connected component of the ....

....amount of memory is required to construct and maintain associating data structures. The hybrid approach presented by the seed set al..gorithm tends to keep the search data structures small. Several algorithms for seed set generation and spatial contour propagation, also called mesh propagation [7, 6], are described in [3] The spatial contour propagation traces and constructs an isosurface component from a single cell by iterating breadth first traversal through the face adjacencies and triangulating until the whole connected surface is constructed. Other marching methods and level sets were ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. In IEEE Transactions on Visualization and Computer Graphics, 1995.

....point of view this important property points out a significant advantage of the hierarchical intersection test compared to other acceleration algorithms for isosurface extraction. Other techniques are for instance the span space methods [40] the k tree method [26] or the extremal graph approach [21]. Without any sophisticated adjustment the expensive preparatory step which comes along with these algorithms has to be invoked on every new interpolation in time. This turns out to a major drawback compared to nearly no extra cost for the hierarchical strategy provided time dependent data is ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundart cell lists. Transactions on Visualization and Computer Graphics, 1(4):319--327, 1995.

....an excellent and thorough review. In Marching Cubes [32] all cells in the volume dataset are searched for isosurface intersection. Techniques avoiding exhaustive scanning include using an octree [56] identifying a collection of seed cells and performing contour propagation from the seed cells [3, 26, 53], NOISE [31] and other nearly optimal isosurface extraction methods [41, 42] The first optimal algorithm was given by Cignoni et al. 11] The first out of core isosurface technique was given by Chiang and Silva [7] They follow the ideas of Cignoni et al. 11] but use the I O optimal interval ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....method is presented, where a particular set of cells is used as seeds to compute iso surfaces. combinatorial data structures: The second step of the contour seeds approach mentioned above consists in using the connections between the cells of the mesh for traversing an iso surface (see also [7, 8]) Since our goal is to generate a series of iso surfaces, it is suitable to exploit the coherency between two consecutive iso surfaces, which is not taken into account by those approaches. Our method can efficiently propagate from one iso surface to the next one, by traversing our combinatorial ....

....between two consecutive iso surfaces, which is not taken into account by those approaches. Our method can efficiently propagate from one iso surface to the next one, by traversing our combinatorial CIEL data structure. This is more efficient than using a sorted list of edges as in [28] In [8], a data structure is introduced, made of a graph connecting the local extrema, where each arc stores the list of cells it traverses. The preprocessing times reported to construct this data structure show that it is not suitable in a dynamic environment, where the scalar fields change. Our ....

[Article contains additional citation context not shown here]

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundart cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995. ISSN 1077-2626.

....is changed by a small delta. One min max list is created for each level of a hierarchical decomposition of the min max search space. The overall complexity remains O(n c ) in the worst case analysis. Extrema graphs Itoh and Koyamada compute a graph of the extrema values in the scalar field [12]. Every connected component of an isocontour is guaranteed to intersect at least one arc in the graph. Isocontours are generated by propagating contours from a seed point detected along these arcs. Noisy data with many extrema will reduce the performance of such a strategy. Livnat et al. note that ....

....One issue is the efficiency of avoiding re computation (recomputing intersection along shared edges of cells) Through marching order and contour propagation, information can be saved more efficiently than in a random order visitation which is caused by some search techniques. Contour propagation [1, 11, 12, 2] is a surface tracking method which is based on continuity of the scalar field, and hence of the isocontours derived from the field. Given a single seed cell on a connected component of a contour, the entire component is traced by breadth first traversal through the face adjacencies. The traversal ....

[Article contains additional citation context not shown here]

ITOH, T., AND KOYAMADA, K. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics 1, 4 (Dec. 1995), 319--327.

....a small fraction of the data set, streaming algorithms that do not avoid reading all of the data can be slower than a demand paging algorithm. The second type of out of core visualization algorithms is indexing algorithms. Many indexing algorithms have been described for isosurface computation [3, 4, 8, 9, 10]. These algorithms precompute an index that identifies the portion of the data that is necessary to compute the requested visualization. For isosurface computation, the index identifies which cells contain portions of the isosurface. One disadvantage of many index algorithms is that their index is ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundart cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....has the property that every iso surface must intersect with at least one cell in the seed set. At run time, given a query, the intersecting cells from the seed set are located and used to propagate to nd all other intersecting cells by exploring the adjacent cells. Extrema Graph Itoh et al. IK95] proposed a method to nd the seed set by building an extrema graph and two sorted lists from the boundary cells. The extrema graph is composed of extrema points and arcs. An extrema point is a cell whose neighbors are all smaller or all bigger in terms of their scalar values, and closer extrema ....

T. Itoh and K. Koyamada. Automatic Isosurface Propagation Using an Extrema Graph And Sorted Boundary Cell Lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319-327, December 1995.

....Next, in the generation phase, depending on the type of cells, one can apply an algorithm to actually generate the isosurface from those active cells. Let N be the total number of cells in the dataset, and K the number of active cells. It is estimated that the average value of K is O#N 2=3 # [9], therefore an exhaustive scanning of all cells in the search phase is found to be inefficient, spending a large portion of time traversing cells that are not active. # Department of Applied Mathematics and Statistics, SUNY Stony Brook, NY 11794 3600; yjc ams.sunysb.edu. Supported in part by NSF ....

....are necessary. Wilhems and Van Gelder [20] propose a method of using an octree to optimize isosurface extraction. This algorithm has worst case time of O#K K log#N=K## (this analysis is presented by Livnat et al. 11] for isosurface queries, once the octree has been built. Itoh and Kayamada [9] propose a method based on identifying a collection of seed cells from which isosurfaces can be propagated by performing local search. Basically, once the seed cells have been identified, they claim to have a nearly O#N 2=3 # expected performance. Livnat et al. 11] estimate the worst case ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....correlation values. Matching two images is therefore equivalent to extracting the maximal surface from the volume. Since the u v d volume is very noisy, simply applying the Marching Cubes algorithm [55] would easily fall into the trap of local maxima. We thus implemented a propagation algorithm [46]. In addition, we make use of the disparity gradient limit [11] which states that d u 1. Use of this constraint in the scanline direction is equivalent to the ordering constraint often used in scanline based algorithms (e.g. 21] by Cox et al. Using it in the direction perpendicular to the ....

T. Itoh and K. Koyamada, "Automatic Isosurface Propagation Using an Extrema Graph and Sorted Boundary Cell Lists," IEEE Trans. Visualization and Computer Graphics, vol. 1, no. 4, pp. 319-327, 1995.

....A method, tensor voting,is described which effectively interpolates smooth structures, detects underlying field discontinuities, and ignores spurious outlier noise inherent in the measurement phase. 1. 1 Related work Many existing articles are related to iso surface extraction (for example, [4, 7, 10, 9, 13]) TheMarchingCubesalgorithm [9] is aclassical solution to this problem. Nielson and Hamann [13] show that the Marching Cubes is not flawless: iso surface with holes may occur owing to anambiguous choiceof four zero crossings. To resolve this ambiguity, they propose an asymptotic decider to ....

....iso surface with holes may occur owing to anambiguous choiceof four zero crossings. To resolve this ambiguity, they propose an asymptotic decider to enforce a deterministic choice to all configurations where such ambiguity can occur. For efficient iso surface extraction, Itoh and Koyamada [7] use complex data structures like extrema graph and sorted boundary cell lists to propagate an iso surface. Livnat et al. 10] propose a span space and the use of kd tree to reduce search time and space by projecting the data set onto a span space. Thirion and Gourdon [19] propose the Marching ....

T. Itoh and K. Koyamada, "Automatic Isosurface Propagation Using an Extrema Graph and Sorted Boundary Cell Lists", IEEE Trans. Visual. Comput. Graphics, vol. 1, no. 4, pp. 319--327, 1995.

....the min and max values 2 on each node in the octree structure (cf. Section 4. 4) The hierarchical approach competes with other efficient isosurface extraction methods which use some efficient presorting [23,55,39] or seed cell algorithms, such as the extremal graph methods by Itoh and Koyamada [32,33]. These approaches are preferable if the data is governed by high frequencies. But in contrast to these approaches, the hierarchical data access, as for instance in 2D, can be combined with an adaptive choice of the desired data resolution. A fast and adaptive visualization of volume data is ....

.... t i (x) From our point of view this important property points out a significant advantage of the hierarchical intersection test compared to other acceleration algorithms for isosurface extraction, including the span space methods [55] the k tree method [39] or the extremal graph approach [32]. Without any sophisticated adjustment 12 the expensive preparatory step which comes along with these algorithms has to be invoked on every new interpolation in time. This turns out to be a major drawback compared to almost no extra cost for the hierarchical strategy, provided time dependent data ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundart cell lists. Transactions on Visualization and Computer Graphics, 1(4):319--327, 1995.

....et al. 29] consider the problem of detecting drainage patterns in geographic terrain. Interrante et al. 22] used ridge and valley detection on 3D surfaces to enhance the shape of transparently rendered surfaces. Extrema graphs were used by Itoh and Koyamada to speed isocontour extraction [23]. A graph containing extreme points and boundary points of a scalar field can be guaranteed to intersect every isocontour at least once, allowing seed points to be generated by searching only the cells contained in the extrema graph. Helman and Hesselink detect vector field topology by classifying ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....so has evidently not appeared in the literature. Seed indices sparse traversal in 3D and 4D Seed indices appear to be more promising as datareducing data structures than value indices. The idea is to store only the subset of the cells from which all other cells may be found (the seed cells) [20], 1] The storage requirements of Itoh are not explicitly noted in the paper, but appear to be in the 10 to 20 range [20] Bajaj in particular shows storage requirements that increase by only 1 to 8 [1] SIGGRAPH 99 48 Version 9 Sep 99 Indexing over pages Ueng built an octree over segments ....

....promising as datareducing data structures than value indices. The idea is to store only the subset of the cells from which all other cells may be found (the seed cells) 20] 1] The storage requirements of Itoh are not explicitly noted in the paper, but appear to be in the 10 to 20 range [20]. Bajaj in particular shows storage requirements that increase by only 1 to 8 [1] SIGGRAPH 99 48 Version 9 Sep 99 Indexing over pages Ueng built an octree over segments of the data, using values at intermediate nodes for reduced resolution volume rendering [43] The system allowed the user to ....

ITOH,T.,AND KOYAMADA, K. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics 1,4 (December 1995), 319--.

....efficiency. Cells, represented by the intervals defined by their extreme values, are grouped at the nodes of a balanced binary tree. For any isovalue query, at most one branch from a node is traversed. An alternate technique is to propagate the isosurface from a set of seed cells. Itoh et al. [8, 9], Bajaj et al. 1] and van Kreveld et al. 19] construct seed sets that contain at least one cell per connected component of each isosurface. The isosurface construction begins at a seed and is traced through neighboring cells using adjacency and intersection information. An algorithm to ....

Takayuki Itoh and Koji Koyamada. Automatic Isosurface Propagation Using an Extrema Graph and Sorted Boundary Cell Lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

....interpolation of gradients at the endpoints of the cell edge where w lies. Purging not active cells The identification of the set of cells crossed by the isosurface entails traversing the whole dataset, even if the isosurface being searched for only crosses a few cells. Many speedup techniques [38,20,14,24] have been proposed in order to avoid the analysis of non active cells. An optimal solution to the search for active cells was proposed by Cignoni et al. in [6,8] The method is based on a data structure called interval tree which encodes a set of intervals on the real line and supports ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an Extrema Graph and sorted boundary cell lists. IEEE ToVCG, 1(4):319--327, 1995.

....fitting approach adopted. As in the previous method, the domain is limited to regular datasets. The empirical computational complexity is very close to that observed for the octree. A completely different geometric approach, based on the use of an extrema graph, is proposed by Itoh and Koyamada [6]. The nodes of the graph are the cells of the volume which hold local extrema data values (local minima or maxima) Each arc of the graph supports a list of the cells connecting its two end nodes. Given an isovalue, an active arc is searched for in the extrema graph. The cells connected to this ....

ITOH, T., AND KOYAMADA, K. Automatic isosurface propagation using an Extrema Graph and sorted boundary cell lists. IEEE Trans. on Vis. and Comp. Graph. 1, 4 (1995), 319--327.

....only store the scalar range of the seeds as intervals in the tree, and a pointer into the mesh. Of course, the seed set must be such that every possible contour of the function passes through at least one seed. Otherwise, contours could be missed. There are a few papers that take this approach [4, 12, 22]. The tracing algorithms to extract a contour from a given seed have been developed before, and they require time linear in the size of the output [2, 11, 12] The objective of this paper is to present new methods for seed set computation. Of a seed set, we require that any possible contour in the ....

....of the function passes through at least one seed. Otherwise, contours could be missed. There are a few papers that take this approach [4, 12, 22] The tracing algorithms to extract a contour from a given seed have been developed before, and they require time linear in the size of the output [2, 11, 12]. The objective of this paper is to present new methods for seed set computation. Of a seed set, we require that any possible contour in the mesh pass through at least one seed. Otherwise we could miss a contour. To construct such a small size seed set, we use a variation of the contour tree, a ....

[Article contains additional citation context not shown here]

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Trans. on Visualization and Computer Graphics, 1:319--327, 1995.

....cells) Notice that the search phase is usually the bottleneck of the entire process, since it searches the 3D dataset and produces 2D data. In fact, letting N be the total number of cells in the dataset and K the number of active cells, it is estimated that the average value of K is O(N 2=3 ) [20]. Therefore an exhaustive scanning of all cells in the search phase is inefficient, and a lot of research efforts have thus focused on developing output sensitive algorithms to speed up the search phase. In the rest of the paper we use N and K as defined above, and M and B to respectively denote ....

....time is needed. This technique does not require the entire dataset to fit into the main memory, but dN=Be disk reads are necessary. Techniques avoiding exhaustive scanning include using an octree [45] identifying a collection of seed cells and performing contour propagation from the seed cells [8, 20, 41], NOISE [24] and other nearly optimal isosurface extraction methods [35, 36] The first optimal isosurface extraction algorithm was given by Cignoni et al. 14] based on the following two ideas. First, by producing for each cell an interval, the active cell searching process is reduced to the ....

T. Itoh and K. Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

No context found.

ITOH, T., AND KOYAMADA, K. 1995. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics 1, 4 (Dec.), 319--327.

No context found.

T Itoh and K Koyamada. Automatic isosurface propagation using an extrema graph and sorted boundary cell lists. IEEE Transactions on Visualization and Computer Graphics, 1(4):319--327, December 1995.

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