Retrieving landscape images using scene structural matrix

Abstract

In this paper, we present Scene Structural Matrix (SSM) and apply it to the retrieval of landscape images. The SSM captures the overall structural characteristics of the scene by indexing the geometric features of the image. A binary image tree (bintree) is used to partition the image and from which we derive multi-resolution geometric structural descriptors of the image. It is shown that SSM is particularly effective in retrieving images with strong structural features, such as landscape photographs. We show that SSM is robust against spatial and spectral distortions thus making it superior to current state of the art techniques such as color correlogram in certain applications. We will also show that images retrieved by the SSM are more relevant than those returned by color correlogram and color histogram. © 2001 Springer-Verlag Berlin Heidelberg.