W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(3):224--251, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....that find one to one correspondences between graph nodes. Shapiro and Haralick [24] proposed a matching algorithm based on comparing weighted primitives (weighted attributes and weighted relation tuples) using a normalized distance for each primitive property that is inexactly matched. Kim and Kak [16] used a combination of discrete relaxation and bipartite matching in model based 3 D object recognition. Pellilo et al. 19] devised a quadratic programming framework for matching association graphs using a maximal clique reformulation, while Gold and Rangarajan [10] used graduated assignment for ....

W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(3):22z1251, 1991.

....is done using maximum clique finding in a 1717 relational graph. A high level strategy uses clusters of features to generate hypotheses, which are then compared with complementary features in a low level verification and parameter refinement step. Bipartite Matching and Relaxation Labeling. [Kim and Kak 1991] used bipartite matching together with discrete relaxation to perform recognition of 3D objects from bin parts using the output of an structured light scanner. This represents to our knowledge one of the first attempts to use the technique of bipartite matching in computer vision. Each object is ....

....to an instance of the assignment problem in quadratic time. He then uses a suboptimal strategy based on constraining the set of feasible matchings by imposing a bounded error model, and presents an algorithm to find a matching and a similarity transformation in average quadratic time. 84 [Kim and Kak 1991] made use of bipartite matching together with discrete relaxation to perform recognition of objects from bin parts using the output of an structured light scanner producing range depth data. Each object is represented by an attributed graph whose nodes are surface features, and whose arcs are ....

Kim, W.Y. and Kak, A.C. 3D Object Recognition Using Bipartite Matching Embedded in Discrete Relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(3): 224---251, 1991.

....recognition systems can also be classified into two categories according to how they deal with multi object scenes. In the first category, systems assume that the input scene consists of either a single object [6, 17, 18, 19] or several objects that can be correctly separated from each other [20, 21]. This assumption enables the utilization of scene perceptual information for improving performance. However, reliance on such an assumption limits the applicability of these systems in complex scenes, where it may be difficult to correctly segment a scene into objects. In the second category, ....

W. Kim and A. C. Kak. 3-D object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. and Mach. Intell., 13(3):224--251, 1991.

....load on subsequent hypothesis generation and verification modules, since only consistent model scene feature pairs have to be considered. Furthermore, LCE can eliminate many erroneous model objects efficiently, without explicitly generating or verifying any object pose hypotheses (cf. [7]) The remainder of the paper is organized as follows. The next section defines the problem. Construction of the CSP is discussed in Section 3. Section 4 presents experimental results, and a number of conclusions are drawn in Section 5. 2 Formulation of the Problem A general CSP is represented ....

W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. and Mach. Intell., 13(3):224--251, 1991.

.... of a node with its neighboring nodes is fundamental to a graph, relaxation based processing extends naturally to computations over graphs using labels as constraints [27] Kitchen and Rosenfeld [19] describe a Discrete Relaxation Labeling scheme for matching relational structures and Kim and Kak [18] have proposed a scheme using bipartite matching embedded in discrete relaxation for matching relational structures for 3 D object recognition. In all cases, the solution found is not guaranteed to be optimal, and Simulated Annealing [1] may be used to increase the likelihood of a globally optimal ....

Kim, W.-Y., and Kak, A. C. 3-D object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence 13, 3 (Mar. 1991), 224--251.

....of the solutions they find, but they are prone to combinatorial explosion in time and in space. It is known that NP complete problems can be addressed by stochastic optimization algorithms with polynomial time complexity such as simulated annealing, genetic algorithms or continuous relaxation [Hae90, KU88, HS88, KK91, CH81]. However these methods do no longer guarantee to find an optimal solution. In practice, most applications have to deal with noisy data and therefore cannot rely solely on exact matching. The problem of inexact graph matching can be defined in various ways and the search algorithms are strongly ....

W.Y. Kim and A.C. Kak. 3-d object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on pattern analysis and machine intelligence PAMI, 13:224--251, 1991.

....robust object localization can be achieved. The problem of 3D object recognition 1 has received significant attention during the last two decades (e.g. see surveys [1, 2, 3, 4] Most 3D object recognition systems rely on a single type of sensory data such as vision [5, 6, 7, 8, 9, 10] range [1, 11, 12, 13], or touch [14, 15, 16] Thus, they are unsuitable for utilizing various types of sensory data which may be readily available in some tasks, such as ours, in order to improve efficiency and robustness. There have been few efforts for integrating visual and tactile data in the context of 3D object ....

W. Kim and A. C. Kak. 3-D object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. and Mach. Intell., 13(3):224--251, 1991.

....load on the hypothesis verification module. Furthermore, using LCE, many erroneous model objects can be completely eliminated (in the case of LCE failure) without either generating or verifying any hypotheses. An approach that is closely related to ours is the one presented by Kim and Kak [15] to recognize 3 D objects from range data. In their work, an input scene and model objects are all represented by attributed graphs. Rather than performing extensive graph matching between the scene graph and every model one, they first apply bipartite matching in order to check for the ....

W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. and Mach. Intell., 13(3):224--251, 1991.

....any possibility of including a given pair of nodes in the final clique. Hence it is somewhat of a devastating operation since it is performed first, during the formation of A g . 1.2. 2 Relaxation Labeling Techniques Widespread effort has been focused on the method of relaxation labeling [20] [21]. This approach describes the compatibility of a given labeling using continuous values, rather than a discrete assignment. Compatibility is based on the expected occurence of a given label and on the consistency of neighboring labels. A cost function is defined that describes the support for a ....

W-Y. Kim and A. C. Kak. 3-d object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. Machine Intell., 13(3):224--251, 1991.

.... by the experimental results in the next section) In contrast, in the alignment case, as well as all the other recognition approaches, the detailed comparison between scene and model features involves much more computations, especially when feature uncertainty is taken into consideration (e.g. [9, 19]) 2. The worst case LCE complexity, O(SM 2 ) is never reached in many cases. In fact, the complexity of eliminating an erroneous structure object pair ranges from O(M 2 ) to O(SM 2 ) since LCE stops once it encounters a null constraint (refer to the LCE algorithm in Section 2.2) By ....

....a 2 D line drawing, assuming a generic class of objects, such as polyhedra [4] The CSP LCE framework has been much less commonly used in 3 D object recognition systems, which generally adopt one of the abovementioned three approaches. A notable exception is the system presented by Kim and Kak [19] for recognizing 3 D objects from range data. In their alignment based system, discrete relaxation is used to reduce the number of generated hypotheses. Conceptually, our work can be viewed as a generalization of the one in [19] in at least two aspects: 1) it integrates 2 D and 3 D data, not only ....

[Article contains additional citation context not shown here]

W. Kim and A. C. Kak. 3-D object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. on Pattern Anal. and Mach. Intell., 13(3):224--251, 1991.

.... in occluded environments [10] the INGEN system for generic object recognition [60] the MULTI HASH system for fast 3 D object recognition [23] and the BONSAI system for object recognition using constrained search [20] Other relevant work on 3 D object recognition from range data includes [17, 19, 27, 38, 57]. Systems that have been demonstrated to recognize 3 D objects using principles of binocular stereo and other multi camera systems with various degrees of effectiveness include [18, 30, 33, 53] A variation on the idea of using multiple 2 D images for recognizing a 3 D object consists of ....

....violating arcs until we find a bipartite match. A sequential search for these unacceptable links between scene and model nodes can become combinatorially extensive, but can be replaced by parallel techniques to make the computation feasible. One way of doing this is by discrete relaxation [38]. In general, relaxation in the computer vision context refers to a manner of iterative processing over a cellular structure in which decisions for each cell are made purely locally but subject to contents in the neighboring cells. Since the connection of a node to its neighboring nodes is ....

[Article contains additional citation context not shown here]

W. Y. Kim and A. C. Kak, 3D Object Recognition Using Bipartite Matching Embedded in Discrete Relaxation, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 13, No. 3, March 1991, pp. 224-251.

No context found.

W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(3):224--251, 1991.

No context found.

W. Kim and A. C. Kak. 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE 13(3):224--251, 1991.

No context found.

Kim, W., and Kak, A., 3D object recognition using bipartite matching embedded in discrete relaxation. IEEE PAMI, 13(3), pp. 224-251, March 1991.

No context found.

W. Y. Kim and A. C. Kak, 3-D object recognition using bipartite matching embedded in discrete relaxation, IEEE Trans. Pattern Anal Mach. Iraell. 13(3), 224-251 (March 1991).

No context found.

W. Kim, A.C. Kak, 3D object recognition using bipartite matching embedded in discrete relaxation, IEEE Trans. Pattern Anal. Mach. Intell. 13 (3) (1991) 224}251.

No context found.

W.-Y. Kim and A. C. Kak. 3-D object recognition using bipartite matching embedded in discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(3):224--251, 1991.

No context found.

W. Kim & A.C. Kak. 3-d object recognition using bipartite matching embedded in discrete relaxation. IEEE Trans. Pattern Analysis and Machine Intelligence, 13(3):224-- 251, March 1991.

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