Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Evaluation of spatial similarity methods for image retrieval

Petrakis Evripidis, Kostas Georgiadis

Full record


URI: http://purl.tuc.gr/dl/dias/C3C066FA-3D37-4598-BEDF-AFB5C2235E2A
Year 2000
Type of Item Conference Poster
License
Details
Bibliographic Citation Euripides Petrakis and Kostas Georgiadis. (2000, Sept.). Evaluation of Spatial Similarity Methods for Image Retrieval. Presented at Signal Processing, Communications and Computer Science (CSCC' 2000). [Online].Available: http://www.intelligence.tuc.gr/~petrakis/publications/pege00.pdf
Appears in Collections

Summary

Similarity retrieval by spatial content (i.e., using multiple objects and their interelationships) in Image DataBases (IDBs) is still an open problem andhas received considerable attention in the literature. In this work, we focus our attention on “queries by example” image and we study methods for answeringsuch queries including the well accepted “editing distance” on Attributed Relational Graphs (ARGs) and the “Hungarian Method” for graph matching. Weevaluate the time responses and the accuracy of these methods using a database of synthetic by realistic medical images. Our experiments indicate that the ARGediting distance, although the slowest, is the most accurate method.

Services

Statistics