Flexible syntactic matching of curves and its application to automatic hierarchical classification of silhouettes

Gdalyahu, Yoram; Weinshall, Daphna

doi:10.1109/34.817410

Cited by 197 publications

(147 citation statements)

References 51 publications

Supporting

Mentioning

145

Contrasting

Unclassified

Order By: Relevance

“…As pointed out by Hummel [21], shape descriptors are particularly suited for achieving generalization: many categories are characterized by their shape rather than by color or texture. For this reason, many authors have used shape representations [1,3,7,15,14,19,20,25,32].…”

Section: Related Workmentioning

confidence: 99%

From Images to Shape Models for Object Detection

2009

View full text Add to dashboard Cite

We present an object class detection approach which fully integrates the complementary strengths offered by shape matchers. Like an object detector, it can learn class models directly from images, and localize novel instances in the presence of intra-class variations, clutter, and scale changes. Like a shape matcher, it finds the boundaries of objects, rather than just their bounding-boxes. This is made possible by a novel technique for learning a shape model of an object class given images of example instances. Furthermore, we also integrate Hough-style voting with a non-rigid point matching algorithm to localize the model in cluttered images. As demonstrated by an extensive evaluation, our method can localize object boundaries accurately, while needing no segmented examples for training (only bounding-boxes) Key-words: object detection, shape matching, contour features * This research was supported by the EADS foundation, INRIA and CNRS. Vittorio Ferrari was funded by a postdoctoral fellowship of the EADS foundation. D'images au modèles de forme pour la détection d'objetsRésumé : Nous proposons une nouvelle approche pour la détection de catégories d'objets qui intègre les avantages des méthodes de mise en correspondance de formes. D'une part, comme les détecteurs d'objets, notre approche est capable d'apprendre les modèles de classe directement à partir des images et de localiser de nouvelles instances malgré des variations intra-classe, des changements d'échelle ou la présence d'un fond encombré. D'autre part, tout comme les méthodes de mise en correspondance de formes, elle ne fournit pas simplement une boite englobante, mais parvient à extraire les frontières des objets. Ceci est possible grâce à une nouvelle technique d'apprentissage capable de construire le modèle de forme d'une catégorie d'objets, étant donné des images contenant des instances de cette catégorie. Afin de localiser le modèle dans des images test, nous combinons un algorithme de vote dans un espace de Hough à un algorithme de mise en correspondance de points pour les formes non rigides. Les résultats expérimentaux montrent que les frontières des objets sont localisées avec précision sans utiliser d'images d'apprentissage segmentées, les boites englobants seules suffisent.

show abstract

Section: Related Workmentioning

confidence: 99%

From Images to Shape Models for Object Detection

2009

View full text Add to dashboard Cite

show abstract

“…The shape correspondence can be defined as matching from the set of landmarks on one shape to that on the next (Gdalyahu and Weinshall, 1999;Petrakis et al, 2002). In (Xie et al, 2008), the authors developed a mechanism to generate coarse segment matching between different instances of an object, based on representative skeletal features.…”

Section: Eświerczmentioning

confidence: 99%

“…Matching is a generic operation in pattern recognition which is used to determine the similarity between two entities (points, curves, or shapes) of the same type (Basri et al, 1998;Gdalyahu and Weinshall, 1999;Latecki and Lakamper, 2000;Liu and Srinath, 1990;Umeyama, 1993;Younes, 1999). Features which are used for shape description can be very different, for example, algebraic moments, area, circularity, eccentricity, compactness, major axis orientation, Euler number, concavity tree, shape numbers (Jain et al, 2000;Zhang and Lu, 2003).…”

Section: Eświerczmentioning

confidence: 99%

Classification in the Gabor time-frequency domain of non-stationary signals embedded in heavy noise with unknown statistical distribution

Swiercz¹

2010

International Journal of Applied Mathematics and Computer Science

View full text Add to dashboard Cite

A new supervised classification algorithm of a heavily distorted pattern (shape) obtained from noisy observations of nonstationary signals is proposed in the paper. Based on the Gabor transform of 1-D non-stationary signals, 2-D shapes of signals are formulated and the classification formula is developed using the pattern matching idea, which is the simplest case of a pattern recognition task. In the pattern matching problem, where a set of known patterns creates predefined classes, classification relies on assigning the examined pattern to one of the classes. Classical formulation of a Bayes decision rule requires a priori knowledge about statistical features characterising each class, which are rarely known in practice. In the proposed algorithm, the necessity of the statistical approach is avoided, especially since the probability distribution of noise is unknown. In the algorithm, the concept of discriminant functions, represented by Frobenius inner products, is used. The classification rule relies on the choice of the class corresponding to the max discriminant function. Computer simulation results are given to demonstrate the effectiveness of the new classification algorithm. It is shown that the proposed approach is able to correctly classify signals which are embedded in noise with a very low SNR ratio. One of the goals here is to develop a pattern recognition algorithm as the best possible way to automatically make decisions. All simulations have been performed in Matlab. The proposed algorithm can be applied to non-stationary frequency modulated signal classification and non-stationary signal recognition.

show abstract

“…While efficient matching algorithms have been developed based on dynamic programming (cf. [9]), the integration of the resulting shape distances with statistical learning of shapes is still an open problem. Secondly, explicit boundary representations are typically constrained to a fixed topology.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear Dynamical Shape Priors for Level Set Segmentation

Cremers

2008

J Sci Comput

View full text Add to dashboard Cite

The introduction of statistical shape knowledge into level set based segmentation methods was shown to improve the segmentation of familiar structures in the presence of noise, clutter or partial occlusions. While most work has been focused on shape priors which are constant in time, it is clear that when tracking deformable shapes certain silhouettes may become more or less likely over time. In fact, the deformations of familiar objects such as the silhouettes of a walking person are often characterized by pronounced temporal correlations.In this paper, we propose a nonlinear dynamical shape prior for level set based image segmentation. Specifically, we propose to approximate the temporal evolution of the eigenmodes of the level set function by means of a mixture of autoregressive models. We detail how such shape priors "with memory" can be integrated into a variational framework for level set segmentation. As an application, we experimentally validate that the nonlinear dynamical prior drastically improves the tracking of a person walking in different directions, despite large amounts of clutter and noise.

show abstract

Flexible syntactic matching of curves and its application to automatic hierarchical classification of silhouettes

Cited by 197 publications

References 51 publications

From Images to Shape Models for Object Detection

From Images to Shape Models for Object Detection

Classification in the Gabor time-frequency domain of non-stationary signals embedded in heavy noise with unknown statistical distribution

Nonlinear Dynamical Shape Priors for Level Set Segmentation

Contact Info

Product

Resources

About