The Adaptive Subspace Map for Image Description and Image Database Retrieval

In this paper, a mixture-of-subspaces model is proposed to describe images. Images or image patches, when translated, rotated or scaled, lie in low-dimensional subspaces of the high-dimensional space spanned by the grey values. These manifolds can locally be approximated by a linear subspace. The adaptive subspace map is a method to learn such a mixture-of-subspaces from the data. Due to its general nature, various clustering and subspace-finding algorithms can be used. If the adaptive subspace map is trained on data extracted from images, a description of the image content is obtained, which can then be used for various classification and clustering problems. Here, the method is applied to an image database retrieval problem and an object image classification problem, and is shown to give promising results.