Image Cloud Visualization Overview

Image Cloud is a Visualization Application that allows the user to navigate through a set of images in a new way. Instead of navigating linearly through a list of images, image cloud displays the list of images as a kind of "cloud" where images, that contain colors distributions that are similar are displayed closely to each other. This means, images that contain similar places, persons or other objects will be displayed in close proximity within the integrated navigator, which makes finding pictures that contain a similar scenery very easy.

Clustering overview

The integrated clustering engine can process either images from the local harddisk or images returned by the integrated flickr search. In addition to navigating through a set of images, the application also visualizes the computed signatures, and displays information about the location of the signature bins in color space. Once a set of signatures has been computed the application allows the user to store it on the harddisk for later use.

Features

Additional features include the ability of the application to generate distance queries that will display only a subset of the images, navigational features and the abitlity to navigate directly to the images source.

Additional information

This project has been created on the basis of the work described in  [1].

The project was created within the class "Visualization 2" on the UT Vienna by Sebastian Sippl

Project files

The project installer can be found here (Use the interactive Help screen for a detailed description of the programs features)

Project source code can be found here (winrar archive)

Source code documentation can be found here.

References:

[1] A Metric for Distributions with Applications to Image Databases, Rubner et al.