A texture based multispectral images indexing

Content-based retrieval based on texture descriptors is very useful in the case of remote sensing databases. In this paper, an attention was devoted to the case of partial request. First, texture-based features are extracted with Gabor filter. A fusion scheme of textures features allowed taking into account the texture contribution of each band in the case of multispectral images. The obtained texture features are then used to segment images database in Regions Of Interest (ROI). Finally, the features of the texture regions are used as indexes for image database retrieval.