S.Dickinson, A.Pentland, A.Rosenfeld. Shape recovery using distributed aspect matching. PAMI, 14(2):174-198, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....on geometric relationships. They are inspired by works on perceptual grouping ( 10] 11] The spatial arrangements of the CCPs are studied, and those that fit some criteria based on geometric relationships are considered as parts. Examples are PARVO [12] and the system developed by Dickinson [13]. PARVO uses symmetry and junctions to group CCPs into parts. Junctions give the clues about the structure of the object and ensure robust grouping. Dickinson system s groups CCPs in many stages. First, arrangements of CCPs are searched. Then, these arrangements of CCPs are grouped into faces, the ....

Dickinson, S.J., A.P. Pentland, and A. Rosenfled, 3-D Shape Recovery Using Distributed Aspect Matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1992. 14(2): pp. 174-198.

....our likelihood measure can be incorporated into Freeman s Bayesian probabilistic scheme by taking the role of conditional image probability. Measuring likelihood: View likelihood of angles was com puted using numerical simulations by Ben Arie [3] and Burns et al. 6] Dickinson et al. [8] empirically found the more likely views of particular objects decomposed into geons. In this earlier work, the analysis of likelihood was carried out for simple image measurements: either discrete (qualitative) or 1 dimensional (angles) Note, however, that the general problem requires the ....

....(angles) Note, however, that the general problem requires the numerical estimation of likelihood when the image measurements change continuously with the viewing parameters; this computation is harder, as it requires the numerical estimation of limits. Thus the simula tion work described in [3, 6, 8] cannot be readily gen eralized to compute view likelihood of general objects. Below we provide a simple expression which can be used to numerically estimate the stability and likelihood profiles of general objects, and identify the most likely and stable views of any object. Measuring ....

S. J. Dickinson, A. P. Pentland, and A. Rosenreid. 3- D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174-198, 1992.

....can be related to each other in several ways. The most famous system in this area is JIM and its successors proposed by Biederman and his colleagues [2] 8] JIM uses geons as geometric primitives, but other systems have been proposed in the computer vision community that use superquadrics [18] [3]. On the other hand several view based theories of object recognition have been developed following the key idea of a global match between the perceived image and some image like views stored in our long term memory. Several vision systems have been proposed in literature, each with its own ....

S.J. Dickinson, A.P. Pentland, and A. Rosenfeld. 3-D Shape Recovery Using Distributed Aspect Matching. IEEE Trans. on Pattern Analysis and Machine Intelligence, 14(2):174--198, 1992.

....of the projection in the image. In general, optimization is done using a deformable volumetric primitive model like superquadrics [2] The second approach studies the spatial arrangements of the CCPs forming the contour of the 2D parts and optionally the interior CCPs enclosed by this contour [3 5]. A set of inference rules based on these arrangements are associated with each volumetric primitive the system can infer. The rules are then applied onto the 2D parts in the image to obtain the volumetric primitive hypotheses. The main advantage of a model fitting approach is its ability to ....

Dickinson, S.J., A.P. Pentland, and A. Rosenfled, 3D Shape Recovery Using Distributed Aspect Matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1992.14(2): pp. 174-198.

....that are able to describe a wide range of shapes. These primitive, called geons, are a set of shape primitives that may be used (through parametrization) to construct a wide variety of single part objects, and through composition to build more complex multi part objects. The OPTICA system [Dickinson et al. 1992] operates by using an aspect hierarchy that describes the topology of a distinct view (or aspect) of one or more of the geon primitives. Each aspect may be a projection of a number of geons, and each component of the hierarchy may be a part of more than one aspect. An associated probability matrix ....

S.J. Dickinson, A.P. Pentland, and A. Rosenfeld. 3D Shape Recovery using Distributed Aspect Matching, IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174--198, February 1992.

....parts and the constraint based reasoning for hypotheses veri cation and the derivation of new image features. 3.3. 1 Image Modeling by Parameterized View Hierarchies To generate image models of building parts and aggregates, we employ a modi ed version of the aspect hierarchies proposed by [9]. Aspect hierarchies describe qualitative image models of volumetric primitives in terms of aspects. Each aspect is decomposed into a hierarchical feature based description, to facilitate the recognition of primitives even in the case of partial occlusion by the detection of its visible features. ....

S. J. Dickinson, A. P. Pentland, and A. Rosenfeld. 3-D Shape Recovery Using Distributed Aspect Matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174-198, 1992.

....notation hereon. y Currently with the Image Analysis Systems Group, Jet Propulsion Laboratory, MS 168 522, 4800 Oak Grove Drive, Pasadena, CA 91109. have addressed the problem using 2 D image contours as input (e.g. 14, 6, 21, 29, 1, 31] while others have used range data as input (e.g. [19, 28, 18, 7, 25]) Due to the difficulty of the problem, assumptions are usually made, restricting the type of objects viewed to be either a restricted class of Generalized Cylinders (GCs) or other basic components (e.g. geons [3] or to possess certain properties such as symmetries. Several researchers have ....

....have suggested superquadrics as a tool for shape description [19, 27, 5, 9] Most of the research in this area has concentrated on finding the best superquadric to fit the given data. The issue of segmentation of the shape into the parts is ignored in most work in this area. Dickinson et al. [7] address the problem of recognizing compound objects which are composed of a limited set of primitives (a subset of Biederman s geons [3] These primitives can be recovered based on their aspect graph. The assumption in this work, as in most cases where compound objects are handled (e.g. 1, ....

S. J. Dickinson, A. P. Pentland, and A. Rosenfeld. 3D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):173--199, February 1992.

....and search requirements are impractical for objects of modest complexity. Eggert et al. 10] observed that the aspect graph is often based on a level of detail not fully observable in practice and explored a notion of a scale space aspect graph to reduce the number of views. Dickinson et al. [8] constructed a hierarchical aspect graph system based on a set of primitives. Ikeuchi and Kanade [11] use the similarity of feature extracted from 2D views of the object to form a graph structure used in recognition. The theory which defines the aspect graph, i.e. view stability and view ....

S. Dickinson, A. Pentland, and A. Rosenfeld. 3D shape recovery using distributed aspect matching. PAMI, 14(2):174-- 198, February 1992.

....is the huge amount of aspects of even relatively simple objects. To cope with it, the idea of finite resolution AG has been proposed [7,30] According to another approach, the recognition of a complex object is performed by identifying in its line drawings sub images of simple 3 D primitives [6]. Anyway, a certain unbalance can be remarked between the efforts aimed at computing the aspect graphs and the practical applications reported. The author feels that the full power of the aspect graph idea has not been fully exploited. First, the full information stored in the AG, that is ....

....of the very concept of aspect graph. Second, topologically matching images of an unknown object and stored aspects appears a most straightforward consequence of the topological nature of the aspect graph. However, even if a few recognition algorithms make use of this idea (see for instance [6]) the capabilities of the topological approach have not been thoroughly investigated. UNCORRECTED PROOF S0004 3702(01)00055 8 FLA AID:1786 Vol. P.3 (1 25) ELSGML 2001 01 18 Prn:24 01 2001; 8:25 AIJ1786 by:ML p. 3 A. Laurentini Artificial Intelligence . 3 1 1 ....

[Article contains additional citation context not shown here]

S. Dickinson, A. Pentland, A. Rosenfeld, 3-D shape recovery using distributed aspect matching, IEEE Trans. Patt. Anal. Machine Intell. 14 (2) (1992) 174--198.

No context found.

S.Dickinson, A.Pentland, A.Rosenfeld. Shape recovery using distributed aspect matching. PAMI, 14(2):174-198, 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld. 3-D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174--198, 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld. 3-D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174--198, 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld. 3-D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174--198, 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld, "Shape Recovery Using Distributed Aspect Matching," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 14, no. 2, pp. 174-198, Feb. 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld. 3-D shape recovery using distributed aspect matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 14(2):174--198, 1992.

No context found.

S. Dickinson, A. Pentland, and A. Rosenfeld. Shape Recovery using Distributed Aspect Matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1992.

.... the area of physics based deformable models, various formulations have been proposed ( 25, 13, 17, 16] As powerful as these techniques are, they typically assume the model representation is fixed and known a priori, sometimes imposing a heavy burden on the model initialization recovery process ([6, 5]) Furthermore, a representational gap exists between the coarse, parametric shapes used to model the objects and the point based discretizations used to bind them to the image; no obvious method exists to bridge this gap through the inclusion of other basic geometric primitives, e.g. lines. On ....

S.Dickinson, A.Pentland, A.Rosenfeld. Shape recovery using distributed aspect matching. PAMI, 14(2):174-198, 1992.

.... the area of physics based deformable models, various formulations have been proposed ( 11, 13, 16, 15] As powerful as these techniques are, they typically assume the model representation is fixed and known a priori, sometimes imposing a heavy burden on the model initialization recovery process ([6, 5]) Furthermore, a representational gap exists between the coarse, parametric shapes used to model the objects and the point based discretizations used to bind them to the image; no obvious method exists to bridge this gap through the inclusion of other basic geometric primitives, e.g. lines. On ....

S.Dickinson, A.Pentland, A.Rosenfeld. Shape recovery using distributed aspect matching. PAMI, 14(2):174-198, 1992.

No context found.

S. J. Dickinson, A. P. Pentland, and A. Rosenfeld. 3-d shape recovery using distributed aspect matching. IEEE Pat. Anal. Mach. Intell., 14(2):174--198, 1992.

No context found.

S.J. Dickinson, A.P. Pentland, and A. Rosenfeld. 3D Shape Recovery Using Distributed Aspect Matching. IEEE Transaction on Pattern Analysis and Machine Intelligence, 14(2):130--154, 1992.

No context found.

S. J. Dickinson, A. P. Pentland, and A. Rosenfeld, 3-D shape recovery using distributed aspect matching, IEEE Trans. Pattern Anal. Machine Intell., 14(2):174--198, 1992.

No context found.

S. J. Dickinson, A. P. Pentland, and A. Rosenfeld. 3-d shape recovery using distributed aspect matching. IEEE Pat. Anal. Mach. Intell., 14(2):174--198, 1992.

No context found.

S.J. Dickinson, A.P. Pentland and A. Rosenfeld, "3-D Shape Recovery Using Distributed Aspect Matching, IEEE Transaction on PAMI, Vol. 14, No. 2, Feb. 1992, 174-190.

No context found.

S. Dickinson, A. Pentland and A. Rosenfeld, 3-D shape recovery using distributed aspect matching, IEEE Trans. Pattern Anal. Mach. lntell. 14(2), 174-198 (1992).

No context found.

Dickinson, S.J., A.P. Pentland, and A. Rosenfled, 3D Shape Recovery Using Distributed Aspect Matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1992.14(2): pp. 174-198.

First 50 documents  Next 50

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