Efficient attributed graph matching and its application to image analysis

Bunke, Horst; Messmer, Bruno T.

doi:10.1007/3-540-60298-4_235

Cited by 36 publications

(14 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…These methods fall under structural approaches. They are then represented as Attributed Relational Graphs (ARG) [Bunke and Messmer, 1995;Conte et al, 2004], Region Adjacency Graphs (RAG) [Lladós et al, 2001], constraint networks [Ah-Soon and Tombre, 2001] as well as deformable templates [Valveny and Martí, 2003]. Their common drawback comes from error-prone raster-to-vector conversion.…”

Section: State-of-the-art 121 Symbol Representationsmentioning

confidence: 99%

Symbol recognition using spatial relations

Santosh

Lamiroy

Wendling

2012

Pattern Recognition Letters

View full text Add to dashboard Cite

In this paper, we present a method for symbol recognition based on the spatio-structural description of a 'vocabulary' of extracted visual elementary parts. It is applied to symbols in electrical wiring diagrams. The method consists of first identifying vocabulary elements into different groups based on their types (e.g., circle, corner ). We then compute spatial relations between the possible pairs of labelled vocabulary types which are further used as a basis for building an Attributed Relational Graph that fully describes the symbol. These spatial relations integrate both topology and directional information.The experiments reported in this paper show that this approach, used for recognition, significantly outperforms both structural and signal-based state-of-the-art methods.

show abstract

Section: State-of-the-art 121 Symbol Representationsmentioning

confidence: 99%

Symbol recognition using spatial relations

Santosh

Lamiroy

Wendling

2012

Pattern Recognition Letters

View full text Add to dashboard Cite

show abstract

“…Ein zu klassifizierender Graph H besitzt das Merkmal P , wenn P mit einer geeigneten Abbildung f der Variablen, die die Knoten des zum Produkt gehörigen Graphen repräsentieren, als Teilstruktur in H enthalten ist. Wir wollen in dieser Arbeit eine Funktion f , die eine relationale Struktur in eine andere Struktur abbildet und mit den Relationen verträglich ist, allgemein als Homomorphismus bezeichnen ( [45] Garey and Johnson haben in [13] gezeigt, daß die Subgraphisomorphie, d. h. die Entscheidung der Einbettbarkeit durch Monomorphismen, und die Entscheidung der θ-Subsumtion NP-vollständige Probleme sind (s. auch [19,24,3]). Die Senkung dieser durch Morphismen entstehenden Berechnungskomplexität ist also einer der entscheidenden Punkte beim relationalen Lernen.…”

Section: Produktdarstellungunclassified

Graphbasierte Lernverfahren für relationale Daten

Geibel

Wysotzki

2000

Informatik Forsch Entw

View full text Add to dashboard Cite

Zusammenfassung. Anhand zweier Anwendungsbeispiele aus der Induktiven Logischen Programmierung (ILP) wird die Fragestellung des Erlernens von Klassifikatoren für relational strukturierte Beispiele betrachtet. Es wird auf wesentliche Techniken und Fragen des graphbasierten und logischen Lernens, insbesondere auf Fragen des Aufwands eingegangen. Darauf aufbauend wird das Verfahren Indigo zur effizienten graphentheoretischen Transformation des relationalen in ein merkmalsbasiertes Lernproblem vorgestellt, welches den Einsatz von klassischen Lernverfahren wie CAL3 oder ID3 erlaubt. Am Beispiel von Indigo wird gezeigt, daß spezifisch graphentheoretische Ansätze und Algorithmen einen wertvollen Beitrag zur Behandlung von relationalen Lernproblemen erbringen können, der mit rein logischen Methoden, wie sie derzeit in der ILP untersucht werden, nicht entsprechend erzielt werden kann. Eine Aufgabenstellung dieser Arbeit ist deshalb die Gegenüberstellung von graphbasierten und logischen Lerntechniken am Beispiel des hybriden Lernverfahrens Tritop, welches Grundideen aus der Graphentheorie in einem logischen Gewand verwirklicht. Tritop verwendet die sog. α-Subsumtion, eine Spezialisierung der bekannten θ-Subsumtion, zur Klassifikation von Beispielen. Neben den beiden Lernverfahren gehen wir auf die effiziente Konstruktion von Klassenprototypen mit einem konnektionistischen Verfahren ein.Abstract. In this article, we discuss the problem of learning a classifier for relational data with two application examples from Inductive Logic Programming (ILP). We present important techniques and questions of graph based and logical learning, especially its computational complexity. Based on this discussion, the learning system Indigo is described, which relies on the efficient graph based transformation of the relational learning task into a feature based problem, which can be solved with classical feature based methods like CAL3 or ID3. Using Indigo as an example, it can be seen that graph theoretical approaches can make an important contribution to relational machine learning, that cannot be achieved in the same way with logical methods stemming from the field of ILP. Therefore, one concern of this work is the comparison of graph theoretical and logical learning methods. This is done by describing the hybrid learning system Tritop that uses graph theoretical methods together with a logical notation and ILP concepts like subsumtion of clauses. Tritop uses the so called α-subsumption, a restriction of the well known θ-subsumption. As an extension of the two learning systems, we also discuss the efficient construction of class prototypes using a neual network.

show abstract

“…After detecting their presence, the impression image is decomposed into a set of primitives. To obtain a structural representation of these primitives, an attributed relational graph(ARG) [51,52] is built. An ARG is a directed graph that can be represented as a 3-tuple (V, E, A) where V is the set of vertices, also called nodes, E is the set of relations (edges) and A is the set of attributes.…”

Section: Graph Representationmentioning

confidence: 99%

Computational Methods for the Analysis of Footwear Impression Evidence

Srihari

Tang

2014

Studies in Computational Intelligence

View full text Add to dashboard Cite

Abstract. Impressions of footwear are commonly found in crime scenes. Yet they are not routinely used as evidence due to: (i) the wide variability and quality of impressions, and (ii) the large number of footwear outsole designs which makes their manual comparison time-consuming and difficult. Computational methods hold the promise of better use of footwear evidence in investigations and also in providing assistance for court testimony. This paper begins with a comprehensive survey of existing methods, followed by identifying several gaps in technology. They include methods to improve image quality, computing features for comparison, measuring the degree of similarity, retrieval of closest prints from a database and determining the degree of uncertainty in identification. New algorithms for each of these problems are proposed. An end-to-end system is proposed where : (i) the print is represented by an attribute relational graph of straight edges and ellipses, (ii) a distance measure based on the earth-mover distance, (iii) clustering to speed-up database retrieval, and (iv) uncertainty evaluation based on likelihoods. Retrieval performance of the proposed design with real crime scene images is evaluated and compared to that of previous methods. Suggestions for further work and implications to the justice system are given.

show abstract

Efficient attributed graph matching and its application to image analysis

Cited by 36 publications

References 19 publications

Symbol recognition using spatial relations

Symbol recognition using spatial relations

Graphbasierte Lernverfahren für relationale Daten

Computational Methods for the Analysis of Footwear Impression Evidence

Contact Info

Product

Resources

About