Similarity Retrieval of Images with the SR-Tree

This demonstration uses a Java Applet.
A browser which supports AWT 1.1.5 is required (e.g., HotJava 1.1, Internet Explorer 4.01, Netscape Communicator 4.05 with AWT 1.1.5 ).

About Images

Photos and videos of space shuttles are downloaded from NASA.

Johnson Space Center

Keneddy Space Center

NASA Shuttle Web

images are used.

About Similarity Retrieval

Similarity is measured in terms of color histograms. Munsell color space (hue, saturation, intensity) is used. It is divided into nine subspaces: black, gray, white, and six colors.

Hitograms of four sub-regions, i.e., upper-left, upper-right, lower-left and lower-right, are calculated in order to take account of the composition of images. Four histograms are concatenated to compose a 36-dimensional feature vector. Then, feature vectors are reduced to 20-dimensional vectors with the principal component analysis.

Similarity of images is measured by Euclidean distance among these 20-dimensional vectors.

Back to the page of the SR-Tree