H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Trans. on Pattern Analysis and Machine Intelligence, 10(6):769--782, November 1988.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....corresponding viewpoint. R approximates the object O more or less closely, depending on the viewpoints and on the shape of O itself. This simple reconstruction technique, which holds for any kind of object and for any set of camera position, is called Folume Intersection (VI) see [1] 4] 15] [17], 18] This approach requires the 3D positions of silhouettes and viewpoints. This is easy to obtain for an object on a turntable or for a mobile camera in a fixed environment, but in several practical situations this information is not available. Consider for instance objects as an aircraft, a ....

Noborio et al.:Construction of the octree approximating three-dimensional objects by using multiple views, IEEE Trans. on PAMI, Vol.10, pp.769-782,1988

....the scene, resulting in an approximation that includes the visual hull as well as other scene points. Figure 16.1 shows an example of the volume constructed from three 1D views of a 2D scene. Many algorithms have been developed for constructing volumetric models from a set of silhouette images [1, 5, 9, 31, 32, 33, 37, 38, 42, 43, 44, 53, 54, 55]. Starting from a bounding volume that is known to enclose the entire scene, the volume is discretized into voxels and the task is to create a voxel occupancy description corresponding to the intersection of back projected silhouette cones. The main step in these algorithms is the intersection ....

....is to create a voxel occupancy description corresponding to the intersection of back projected silhouette cones. The main step in these algorithms is the intersection test. Some methods back project the silhouettes, creating an explicit set of cones that are then intersected either in 3D (e.g. [43, 54]) or in 2D after projecting voxels into the 472 ....

H. Noborio, S. Fukada, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769-- 782, 1988.

....the scene, resulting in an approximation that includes the visual hull as well as other scene points. Figure 1.1 shows an example of the volume constructed from three 1D views of a 2D scene. Many algorithms have been developed for constructing volumetric models from a set of silhouette images [1, 5, 9, 31, 32, 33, 37, 38, 42, 43, 44, 53, 54, 55]. Starting from a bounding volume that is known to enclose the entire scene, the 4 Figure 1.1 Reconstruction from three silhouettes. The generalized cones associated with the three images result in a reconstruction that includes the object (black) points in concavities that are not visible from ....

....a voxel occupancy description corresponding to the intersection of back projected silhouette cones. The main step in these algorithms is the intersection test. Some methods back project the silhouettes, creating an explicit set of 3D volume cones that are then intersected in either 3D (e.g. [43, 54]) or 2D after projecting voxels into the images (e.g. 23, 44] Alternatively, each voxel can decide if it is in the intersection by projecting it into all of the images and testing if it is contained in every silhouette [55] To make this scene space traversal more efficient, most methods use ....

H. Noborio, S. Fukada, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769-- 782, 1988.

....reconstruction can be generally classified as surface based approaches or volume based approaches [9,15] Surface based approaches have been reported in the literature [61,12,75,74] in which the surface of the object can be described by a single closed parametric grid. Volume based approaches [45,58,21] to the problem describe the object volumetrically and do not provide direct surface information about the object. Cylindrical and spherical coordinate systems centered in the object are commonly used in surface based approaches for resampling the surface of curved objects. These two coordinate ....

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the Octrees Approximating Three-Dimensional Objects by Using Multiple Views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769--782, November 1988.

....shape cues, it achieved very good results in accuracy and time performance, even as a stand alone module. With the current implementation, an image sequence of 18 views was usually processed in less than one minute on an IBM RS 6000 or an SGI Reality Engine. In contrast to octree based methods (Noborio, et al., 1988), the computing time of this algorithm is independent of the number of contour segments in an image. Therefore, octree algorithms will be fast for objects with simple shapes but will loose speed for complicated surfaces. Minor errors due to an imperfection of the lighting setup and some ....

Noborio, H., Fukuda, S., Arimoto, S., 1988. Construction of the Octree Approximating ThreeDimensional Objects by Using Multiple Views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6), pp. 769-782.

....volume cubes with the object silhouette; a cube projected into the image plane can form a hexagon which can lead to a time consuming intersection test. The approximation of this hexagon by a bounding box as done by several authors can lead to reconstruction errors. Moreover, some authors [1] 12] [10] approximate the object silhouette with a polygon, which can lead to reconstruction errors as well, especially in case of complex silhouettes with holes. The speed of computation also depends on the way the central projection is performed; for that reason the choice of a suitable camera model is ....

....INFORMATION USING PILLAR LIKE VOLUME ELEMENTS A keypoint in performing the method of occluding contours is a proper volume representation, characterized by low complexity and suitability for a fast computation of volume models. One often used representation are octrees [2] 3] 5] 6] 9] [10], 11] 13] which describe the object as a tree of recursively subdivided cubes down to the finest resolution. The volume model is constructed projecting the octree cubes consecutively into the image planes of every viewpoint and testing them for intersection with the object silhouettes. One ....

H. Noborio, S. Fukada, and S. Arimoto, "Construction of the octree approximating three--dimensional objects by multiple views", Trans. on Pattern Analysis and Machine Intelligence, Vol. PAMI--10(6), 1988, pp. 769--782.

....of individual objects are labeled to identify the object correspondance. Through experimental analysis, Potmesil anticipates that sets of 16 to 64 images will be required to produce accurate models. 2. 4 Construction of the octree approximation three dimensional objects by using multiple views [NFA88] Noborio, Fukada and Arimoto describe an algorithm for constructing an octree from several images. These images are acquired from arbitrary viewpoints. The algorithm processes polyhedral cones represented by the B rep model. A polyhedral cones is defined by a viewpoint and a silhouette polygonal ....

....paper. The complexity analysis will be treat in the next section. Table 2 presents the number of objects per scene that algorithms admit. Looking the table we can conclude: 1. Some reconstruction algorithms can handle several objects in the scene [HS85] Pot87] while others admit just one [CA86] [NFA88], LPR89] AV89] SA90] Fra90] Sze93] The first case is useful for dealing with the workspace of a robot. 2. All methods generate a single octree model to represent the whole scene even those that admit several objects. We want to mention that they do this by storing a set of tags to ....

[Article contains additional citation context not shown here]

H. Noborio, S. Fukada, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769--782, 1988.

....Surfaces Volume Topology Surface corres. Free space Unknown space B reps, meshes [6, 9, 11, 13, 20, 35, 44, 43] p p p p p p p OPUS b rep [22] p p p p p p p p Edge or surface patch sets [27, 41, 46] p p p p p p Closed quadrics [15, 28, 33, 37, 38, 42, 45] p p p p p Voxels, occupancy grids [1, 3, 30, 34] p p p p p Space envelope b rep p p p p p p p p p p Table 1: Summary of capabilities and information available in modeling representations. Gibson noted that a human subject s perception of isolated stimuli failed to correlate to the subject s ability to perceive in practical settings [18] He ....

.... objects visual Blais 5 models, single isolated Levine 95 [9] mesh 14 images objects visual Li 2 models, single and multiple Grebbin 94 [30] octtree 16 images isolated objects visual Noborio, Fukuda 4 models, single isolated visual, Arimoto 88 octtree 103 images objects, robot surface to [34] (simulated) workspace octtree error Asada, Kimura, 1 model, general outdoor visual, Taniguchi Shirai height 6 images (miniature plane fit to 92 [3] maps model) grid area error Hoover, Goldgof OPUS 2 images shown, single isolated visual, model Bowyer 95 [22] b rep 218 mentioned objects ....

[Article contains additional citation context not shown here]

H. Noborio, S. Fukuda and S. Arimoto, "Construction of the Octree Approximating Three-Dimensional Objects by Using Multiple Views", in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 10, no. 6, November 1988, pp. 769-781.

....and shown to improve the ability of the method to reconstruct complex shapes in the presence of noise. Algorithms for 2D pixel and 3D voxel data are described and applied to 3D ultrasound data. 1 Introduction Volume intersection methods are a popular way of obtaining shape from silhouette [2, 3, 4, 5, 6, 7]. Sources of silhouette data include laser range data, video and thermal imaging [10] A silhouette can be thought of as a binary valued projection of an object space. A set of silhouettes specifies a bounding volume obtained by back projecting each silhouette and finding the associated ....

....presented for a 3D ultrasound phantom. 2 Related work 2. 1 Shape from silhouette Two types of volume intersection algorithms are known for extracting shape from silhouette, those that employ a slowly changing viewpoint between image frames [4] and those based on a sparse set of distinct views [6]. In both approaches each image defines a cone of rays formed by back projecting the silhouette. The object is described as the volume which lies within all the cones. If every possible viewpoint outside the convex hull of the object were used, the volume that lay within all the cones would be the ....

[Article contains additional citation context not shown here]

H. Noborio, S. Fukuda, and S. Arimoto. "Construction of the octree approximating three-dimensional objects by using multiple views," IEEE Transactions on PAMI, 10(6):769--781, November 1988.

....the conic volume formed by the silhouette and the center of projection is computed for each viewpoint, and the octrees from all of the viewpoints are then intersected. Both approaches project the octree cubes into the image plane to perform the object silhouette intersection. In contrast to this, Noborio et al. 1988] and Srivastava and Ahuja [1990] perform the intersections in 3 space, which eliminates the need for perspective projection. Both of these approaches use a polygonal approximation to the silhouette. Additional research has also been done on building octrees from range data [Chien et al. 1988] and ....

H. Noborio, S. Fukada, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-10(6):769--782, November 1988.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Trans. on Pattern Analysis and Machine Intelligence, 10(6):769--782, November 1988.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto, "Construction of the octree approximating three-dimensional objects by using multiple views," IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 10, no. 6, pp. 769--782, November 1988.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating a threedimensional object by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769--82, 1988.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating a threedimensional object by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, 10(6):769--82, 1988.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-10(6):769--782, 1989.

No context found.

H. Noborio, S. Fukuda, S. Arimoto, Construction of the octree approximating threedimensional objects by using multiple views, IEEE Trans. on Pattern Analysis and Machine Intelligence 10 (6) (1988) 769--782.

No context found.

H. Noborio, S. Fukuda, and S. Arimoto. Construction of the octree approximating three-dimensional objects by using multiple views. IEEE Transactions Pattern Analysis and Machine Intelligence, 10(6):769--782, November 1988.

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