H. Fashandi, J.F. Peters, and S. Ramanna(Canada)
Features, histogram distance, image retrieval, image similarity, L2 norm, similarity measure
The problem considered in this paper is how to use feature histogram distances to measure image similarities. This article introduces an approach to measuring image similarity based on the L2 norm (also called Euclidean norm) of vectors of distances between one or more 1-dimensional image feature histograms. Feature histograms exhibit the distribution of local features inside an image. A comparison of image retrieval results with the proposed approach is given using the ’SIMPLIcity’ image test set . Similarity measures based on a gathering together (concresence) of distances between features histograms in vectors make it possible to compare images at the level L2 norms and obtain a global view of image similarities. The contribution of this paper is a proposed L2 norm distance image similarity measure.
Important Links:
Go Back
