Robust hierarchical indexing based on texture features

In this paper, we present a hierarchical indexing method based on texture c haracterization for image retrieval. The novelty of our contribution is the hierarchical structure of the index: it exploits the multiresolution formu lation of Wavelet Transforms to define a new set of approximated versions o f the images for each level of resolution. On this set, the algorithm extra cts significant signatures by means of statistical correlations; the experi mental results and the analysis of computational complexity have proved tha t the algorithm presents the best performance at the highest level of the i ndexing hierarchy, where the computational complexity is the lowest. Our me thod has been evaluated by the following methodologies: (a) the study of th e computational complexity for signature generation; (b) the comparison wit h analogous methods based on texture analysis by reporting the performance obtained on the same database (Brodatz); and (c) the evaluation of the robu stness of the hierarchical indexing in different color spaces, by querying the database with different versions of the original images obtained by noi se addition (gaussian and scanner acquisition noise and lossy compression d istortion), brightness and contrast enhancement, color and scale adjustment and rotation. Even if our method is designed for texture databases, experi ments show satisfactory results also on a real heterogeneous photographic d atabase. This confirms the possibility of exploiting our method as a low co mputational complexity indexing tool based on texture characterization in a broader system for hierarchical content-based retrieval. (C) 2000 Academic Press.