Application of composite features in content-based remote sensing image retrieval

With increasing huge amount of remote sensing data collected from various airborne and satellite sensors, how to find the wanted image has been one of the key problems in the management of multi-source remote sensing image database. This paper discusses our view of content-based remote sensing image retrieval and our experiences. In this paper, a retrieval approach based on combining texture feature with spatial relation is applied in the similarity retrieval of multi-source remote sensing image database which consists of different resolution, different waveband, different platform, and different temporal remote sensing images. Comparisons about three texture feature extraction methods are also presented. It was indicated by experimental results that the retrieval results obtained from corribined-features has powerful practical merits than the traditional methods such as gray level co-occurrence matrix method and the tree-structured wavelet transform or wavelet packets method.