Yossi Rubner, Leonidas Guibas and Carlo Tomasi,
The Earth Mover's Distance, Multi-Dimensional Scaling, and
Color-Based Image Retrieval,
In Proceedings of the ARPA Image Understanding Workshop,
May 1997.
Yossi Rubner, Leonidas Guibas and Carlo Tomasi,
The Earth Mover's Distance, Multi-Dimensional Scaling, and
Color-Based Image Retrieval,
In Proceedings of the ARPA Image Understanding Workshop,
May 1997.In this paper we present a novel approach to the problem of navigating through a database of color images. We consider the images as points in a metric space in which we wish to move around so as to locate image neighborhoods of interest, based on color information. The database images are mapped to distributions in color space, these distributions are appropriately compressed, and then the distances between all pairs I, J of images are computed based in the work needed to rearrange the mass in the compressed distribution representing I to that of J. We also propose the use of multi-dimensional scaling (MDS) techniques to embed a group of images as points in a two- or three-dimensional Euclidean space so that their distances are preserved as much as possible. Such geometric embeddings allow the user to perceive the dominant axes of variation in the displayed image group. In particular, displays of 2-d MDS embeddings can be used to organize and refine the results of a nearest-neighbor query in a perceptually intuitive way. By iterating this process, the user is able to quickly navigate to the portion of the image space of interest.