Conference Paper

An Experimental Comparison of Feature Extraction and Distance Metrics for Image Retrieval

Conference: XXVIII Conference on Graphics, Patterns and Images (SIBGRAPI WiP 2015)
Location: Salvador, Bahia, Brazil
Date: June 2015

Authors: Ramon Figueiredo Pessoa, William Robson Schwartz, Jefersson Alex dos Santos
Institution: Department of Computer Science, Universidade Federal de Minas Gerais (UFMG), Belo Horizonte, Brazil

Abstract

This paper seeks to do a comparative study of different features and distance metrics in order to analyze the impact of these factors in the process of Content-Based Image Retrieval (CBIR).

One of the main contributions of this work was statistically analyze the impact of distance metrics in the process of image retrieval by content. We also observed statistically the impact of the variability among different classes and also the variability between images of the same image class.

The results showed, for a sample collected, that the variation attributed to the class is approximately 99.85%. This confirms the fact that each algorithm will work best in a given situation.

The comparative study showed the algorithms which had better accuracy rate to recover different image classes (in the dataset analysed) and also presented the reasons that possibly made these algorithms had better accuracy rate.

Resources

BibTeX

@inproceedings{figueiredo2015featureextraction,
  title={An Experimental Comparison of Feature Extraction and Distance Metrics for Image Retrieval},
  author={Figueiredo Pessoa, Ramon and Schwartz, William Robson and dos Santos, Jefersson Alex},
  booktitle={XXVIII Conference on Graphics, Patterns and Images (SIBGRAPI WiP 2015)},
  year={2015},
  month={June},
  address={Salvador, Bahia, Brazil},
  organization={IEEE}
}
June 2015