A graph matching method and a graph matching distance based on subgraph assignments

Raveaux, Romain; Burie, Jean-Christophe; Ogier, Jean-Marc

doi:10.1016/j.patrec.2009.10.011

Cited by 41 publications

(25 citation statements)

References 52 publications

(55 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We undertook three series of experiments to evaluate the robustness and the accuracy of our similarity measure. We compared our approach with the initial framework based on local structures comparison presented in [15] and [14] and described in Section 1. This approach is denoted L Sim and do not use angles in the local structures of the vertices.…”

Section: Discussionmentioning

confidence: 99%

“…In this context, [15] proposes a polynomial-time framework based on a fast bipartite assignment procedure mapping nodes and their local structures of one graph to nodes and their local structures of another graph. Similar approaches are also presented in [14]. Graph invariants have been efficiently used to solve the graph comparison problem in general and the graph isomorphism problem in particular.…”

Section: Background An Motivationmentioning

confidence: 99%

“…The second level of the proposed approach is a local similarity measure where the graphs are compared based on their local descriptions using edit operations as in [15] and [14]. However, we extended local descriptions by considering angles degree and labels of the edges.…”

Section: A Multi-level Graph Comparison Approach For Kite Recognitionmentioning

confidence: 99%

“…However, we extended local descriptions by considering angles degree and labels of the edges. We denote the obtained similarity measure by GeoL Sim for Geometric Local similarity measure to distinguish it from L Sim the local similarity measure of [15] and [14] that do not uses angles. Thus, with GeoL Sim each vertex of the compared graphs is represented by a signature, i.e.…”

Section: A Multi-level Graph Comparison Approach For Kite Recognitionmentioning

confidence: 99%

See 3 more Smart Citations

Kite Recognition by Means of Graph Matching

Madi

Kheddouci

Bichot

et al. 2015

Graph-Based Representations in Pattern Recognition

View full text Add to dashboard Cite

Section: Discussionmentioning

confidence: 99%

Section: Background An Motivationmentioning

confidence: 99%

Section: A Multi-level Graph Comparison Approach For Kite Recognitionmentioning

confidence: 99%

Section: A Multi-level Graph Comparison Approach For Kite Recognitionmentioning

confidence: 99%

See 2 more Smart Citations

Kite Recognition by Means of Graph Matching

Madi

Kheddouci

Bichot

et al. 2015

Graph-Based Representations in Pattern Recognition

View full text Add to dashboard Cite

“…Furthermore, suboptimal solutions are computed based on subgraphs assignment [17]. For geometric graphs, measuring the similarity between two subgraphs with spatial constraints is a hard task.…”

Section: Introductionmentioning

confidence: 99%

Efficient geometric graph matching using vertex embedding

Armiti

Gertz

2013

Proceedings of the 21st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems

View full text Add to dashboard Cite

For many applications such as road network analysis and image processing, it is critical to study spatial properties of objects in addition to object relationships. Geometric graphs provide a suitable modeling framework for such applications, where vertices are located in some 2D space. For applications where the similarity between the structures of different graphs plays an important role, typically, inexact graph matching algorithms are employed. However, graph matching algorithms face many problems such as scalability with respect to graph size and less tolerance to changes in graph structure or labels.In this paper, we propose a solution to the problem of inexact graph matching for geometric graphs in the 2D space. Our approach allows to effectively answer subgraph and common subgraph queries for geometric graphs that differ in structure, spatial properties, and labels. Initially, a spatial feature is extracted from each vertex, and string edit distance is used to find the distance between pairs of vertices. To speed up graph matching, we propose vertex embedding into the Euclidean space. Based on this, the distance between two vertices can be computed using the Euclidean distance in constant time. To answer subgraph and common subgraph queries, we introduce an iterative matching algorithm that matches two graphs using their similarity in the Euclidean space. Such an algorithm merges highly similar vertices to create similar connected subgraphs. Using representative geometric graphs extracted from road networks, we show that our approach outperforms existing graph matching approaches in terms of matching quality and runtime.

show abstract

References

2011

Data Management of Protein Interaction Networks

View full text Add to dashboard Cite

A graph matching method and a graph matching distance based on subgraph assignments

Cited by 41 publications

References 52 publications

Kite Recognition by Means of Graph Matching

Kite Recognition by Means of Graph Matching

Efficient geometric graph matching using vertex embedding

References

Contact Info

Product

Resources

About