Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-1/W3, 79-84, 2013
© Author(s) 2013. This work is distributed under
the Creative Commons Attribution 3.0 License.
24 Sep 2013
R. Bahmanyar1, G. Rigoll2, and M. Datcu1 1Munich Aerospace Faculty, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, Germany
2Munich Aerospace Faculty, Institute for Human-Machine Communication, Technical University of Munich, Theresienstr. 90, 80333 Munich, Germany
Keywords: Clustering, Internal cluster indexing, External cluster indexing, Information Retrieval systems, Feature extraction, Earth Observation Abstract. The volume of Earth Observation data is increasing immensely in order of several Terabytes a day. Therefore, to explore and investigate the content of this huge amount of data, developing more sophisticated Content-Based Information Retrieval (CBIR) systems are highly demanded. These systems should be able to not only discover unknown structures behind the data, but also provide relevant results to the users' queries. Since in any retrieval system the images are processed based on a discrete set of their features (i.e., feature descriptors), study and assessment of the structure of feature space, build by different feature descriptors, is of high importance. In this paper, we introduce a clustering-based approach to study the content of image collections. In our approach, we claim that using both internal and external evaluation of clusters for different feature descriptors, helps to understand the structure of feature space. Moreover, the semantic understanding of users about the images also can be assessed. To validate the performance of our approach, we used an annotated Synthetic Aperture Radar (SAR) image collection. Quantitative results besides the visualization of feature space demonstrate the applicability of our approach.
Conference paper (PDF, 2067 KB)

Citation: Bahmanyar, R., Rigoll, G., and Datcu, M.: A CLUSTERING-BASED APPROACH FOR EVALUATION OF EO IMAGE INDEXING, Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-1/W3, 79-84,, 2013.

BibTeX EndNote Reference Manager XML