Yakov Keselman

DePaul University, Chicago, Illinois, United States

Are you Yakov Keselman?

Claim your profile

Publications (11)9.99 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: The mainstream object categorisation community relies heavily on object representations consisting of local image features, due to their ease of recovery and their attractive invariance properties. Object categorisation is therefore formulated as finding, that is, `detecting`, a one-to-one correspondence between image and model features. This assumption breaks down for categories in which two exemplars may not share a single local image feature. Even when objects are represented as more abstract image features, a collection of features at one scale (in one image) may correspond to a single feature at a coarser scale (in the second image). Effective object categorisation therefore requires the ability to match features many-to-many. In this paper, we review our progress on three independent object categorisation problems, each formulated as a graph matching problem and each solving the many-to-many graph matching problem in a different way. First, we explore the problem of learning a shape class prototype from a set of class exemplars which may not share a single local image feature. Next, we explore the problem of matching two graphs in which correspondence exists only at higher levels of abstraction, and describe a low-dimensional, spectral encoding of graph structure that captures the abstract shape of a graph. Finally, we embed graphs into geometric spaces, reducing the many-to-many graph-matching problem to a weighted point matching problem, for which efficient many-to-many matching algorithms exist.
    IET Computer Vision 11/2012; 6(6):500-513. DOI:10.1049/iet-cvi.2012.0030 · 0.76 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Object recognition can be formulated as matching image features to model features. When recognition is exemplar-based, feature correspondence is one-to-one. However, segmentation errors, articulation, scale difference, and within-class deformation can yield image and model features which don’t match one-to-one but rather many-to-many. Adopting a graph-based representation of a set of features, we present a matching algorithm that establishes many-to-many correspondences between the nodes of two noisy, vertex-labeled weighted graphs. Our approach reduces the problem of many-to-many matching of weighted graphs to that of many-to-many matching of weighted point sets in a normed vector space. This is accomplished by embedding the initial weighted graphs into a normed vector space with low distortion using a novel embedding technique based on a spherical encoding of graph structure. Many-to-many vector correspondences established by the Earth Mover’s Distance framework are mapped back into many-to-many correspondences between graph nodes. Empirical evaluation of the algorithm on an extensive set of recognition trials, including a comparison with two competing graph matching approaches, demonstrates both the robustness and efficacy of the overall approach.
    International Journal of Computer Vision 08/2006; 69(2):203-222. DOI:10.1007/s11263-006-6993-y · 3.53 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: One of the bottlenecks of current recognition (and graph matching) systems is their assumption of one-to-one feature (node) correspondence. This assumption breaks down in the generic object recognition task where, for example, a collection of features at one scale (in one image) may correspond to a single feature at a coarser scale (in the second image). Generic object recognition therefore requires the ability to match features many-to-many. In this paper, we will review our progress on three independent object recognition problems, each formulated as a graph matching problem and each solving the many-to-many matching problem in a different way. First, we explore the problem of learning a 2-D shape class prototype (represented as a graph) from a set of object exemplars (also represented as graphs) belonging to the class, in which there may be no one-to-one correspondence among extracted features. Next, we define a low-dimensional, spectral encoding of graph structure and use it to match entire subgraphs whose size can be different. Finally, in very recent work, we embed graphs into geometric spaces, reducing the many-to-many graph matching problem to a weighted point matching problem, for which efficient many-to-many matching algorithms exist.
    Cognitive Vision Systems, Sampling the Spectrum of Approaches [based on a Dagstuhl seminar]; 01/2006
  • Source
    Yakov Keselman, Sven Dickinson
    [Show abstract] [Hide abstract]
    ABSTRACT: The recognition community has typically avoided bridging the representational gap between traditional, low-level image features and generic models. Instead, the gap has been artificially eliminated by either bringing the image closer to the models using simple scenes containing idealized, textureless objects or by bringing the models closer to the images using 3D CAD model templates or 2D appearance model templates. In this paper, we attempt to bridge the representational gap for the domain of model acquisition. Specifically, we address the problem of automatically acquiring a generic 2D view-based class model from a set of images, each containing an exemplar object belonging to that class. We introduce a novel graph-theoretical formulation of the problem in which we search for the lowest common abstraction among a set of lattices, each representing the space of all possible region groupings in a region adjacency graph representation of an input image. The problem is intractable and we present a shortest path-based approximation algorithm to yield an efficient solution. We demonstrate the approach on real imagery.
    IEEE Transactions on Pattern Analysis and Machine Intelligence 08/2005; 27(7):1141-56. DOI:10.1109/TPAMI.2005.139 · 5.69 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Object recognition systems have their roots in the AI com- munity, and originally addressed the problem of object categorization. These early systems, however, were limited by their inability to bridge the representational gap between low-level image features and high-level object models, hindered by the assumption of one-to-one correspondence between image and model features. Over the next thirty years, the main- stream recognition community moved steadily in the direction of exem- plar recognition while narrowing the representational gap. The commu- nity is now returning to the categorization problem, and faces the same representational gap as its predecessors did. We review the evolution of object recognition systems and argue that bridging this representa- tional gap requires an ability to match image and model features many- to-many. We review three formulations of the many-to-many matching problem as applied to model acquisition and object recognition.
  • [Show abstract] [Hide abstract]
    ABSTRACT: In recent work, we presented a framework for many-to-many matching of multi-scale feature hierarchies, in which features and their relations were captured in a vertex-labeled, edge-weighted directed graph.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In recent work, we presented a framework for many-to-many matching of multi-scale feature hierarchies, in which features and their re- lations were captured in a vertex-labeled, edge-weighted directed graph. The algorithm was based on a metric-tree representation of labeled graphs and their metric embedding into normed vector spaces, using the embed- ding algorithm of Matou sek (13). However, the method was limited by the fact that two graphs to be matched were typically embedded into vector spaces with dieren t dimensionality. Before the embeddings could be matched, a dimensionality reduction technique (PCA) was required, which was both costly and prone to error. In this paper, we introduce a more ecien t embedding procedure based on a spherical coding of di- rected graphs. The advantage of this novel embedding technique is that it prescribes a single vector space into which both graphs are embedded. This reduces the problem of directed graph matching to the problem of geometric point matching, for which ecien t many-to-many match- ing algorithms exist, such as the Earth Mover's Distance. We apply the approach to the problem of multi-scale, view-based object recognition, in which an image is decomposed into a set of blobs and ridges with automatic scale selection.
    Computer Vision - ECCV 2004, 8th European Conference on Computer Vision, Prague, Czech Republic, May 11-14, 2004. Proceedings, Part I; 01/2004
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Graph matching is an important component in many object recognition algorithms. Although most graph matching algorithms seek a one-to-one correspondence between nodes, it is often the case that a more meaningful correspondence exists between a cluster of nodes in one graph and a cluster of nodes in the other. We present a matching algorithm that establishes many-to-many correspondences between nodes of noisy, vertex-labeled weighted graphs. The algorithm is based on recent developments in efficient low-distortion metric embedding of graphs into normed vector spaces. By embedding weighted graphs into normed vector spaces, we reduce the problem of many-to-many graph matching to that of computing a distribution-based distance measure between graph embeddings. We use a specific measure, the earth mover's distance, to compute distances between sets of weighted vectors. Empirical evaluation of the algorithm on an extensive set of recognition trials demonstrates both the robustness and efficiency of the overall approach.
    Computer Vision and Pattern Recognition, 2003. Proceedings. 2003 IEEE Computer Society Conference on; 07/2003
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Scale-space feature hierarchies can be conveniently representedas graphs, in which edges are directed from coarser features toner features. Consequently, multi-scale feature matching can be formulatedas hierarchical graph matching. Most approaches to graph matchingassume a one-to-one correspondence between nodes (features) which,due to noise, scale discretization, and feature extraction errors, is overlyrestrictive. In general, a subset of features in one hierarchy, representingan ...
    Scale Space Methods in Computer Vision, 4th International Conference, Scale-Space 2003, Isle of Skye, UK, June 10-12, 2003, Proceedings; 01/2003
  • Source
    Yakov Keselman, Sven Dickinson
    [Show abstract] [Hide abstract]
    ABSTRACT: The recognition community has long avoided bridging the representational gap between traditional, low-level image features and generic models. Instead, the gap has been ar- tificially eliminated by either bringing the image closer to the models, using simple scenes containing idealized, tex- tureless objects, or by bringing the models closer to the im- ages, using 3-D CAD model templates or 2-D appearance model templates. In this paper, we begin by examining this trend and track its evolution over the last 30 years. We ar- gue for the need to bridge (not eliminate) this representa- tional gap, and review our recent progress for the domain of model acquisition. Specifically, we address the problem of automatically acquiring a generic 2-D view-based class model from a set of images, each containing an exemplar object belonging to that class. We introduce a novel graph- theoreticalformulationof the problem, anddemonstratethe approach on real imagery.
  • Source
    Craig C Ewert, Yakov Keselman
    [Show abstract] [Hide abstract]
    ABSTRACT: The authors investigate the use of a genetic algorithm to control the evolution of artificial neural networks for the purpose of detecting edges in single-line digitized images. Special-purpose neurons are used in an architecture which is fixed in terms of inter-layer connectivity but free in terms of number of layers, activation functions, and other attributes. Preliminary results indicate that three of the free attributes do have a best form.