Publication Details
Overview
 
 
Iris Vanhamel, Cosmin Mihai, Hichem Sahli, Ioannis Pratikakis
 

Chapter in Book/ Report/ Conference proceeding

Abstract 

A region oriented strategy for content-based image retrieval is presented. It is based on a meaningful segmentation procedure that can provide proper distributions for matching through the Earth Mover's Distance (EMD) as a similarity metric. The segmentation procedure is based on a multiscale watershed-driven algorithm which extracts meaningful regions using a graph-cut algorithm that operates on a multiscale region adjacency graph. In this framework, the proposed robust feature extraction and the many-to-many region matching along with the region weighting for enhancing feature discrimination play a major role. Experimental results demonstrate the performance of the proposed strategy.

Reference