Research On Application Of Hierarchical Index Structure Based On Clustering Analysis In Image Retrieval

Along with the development of information technology, especially the development of Internet technology, multi-media information, including image data, is rapid growing. It becomes more and more necessary to retrieving images faster and efficiently. As traditional method, notes-based image retrieval has been unable to meet the needs. Therefore, the Content-based Image Retrieval (CBIR) has been drawing more and more research attention in the recent years.This paper presents the development of CBIR technology, the key technology of CBIR and evaluation of CBIR. Image feature extraction is the basis of CBIR, and this paper describes the feature of images from the color, shape, texture and relationship between spaces.The traditional CBIR is ordinal retrieval. However, for large-volume and high-dimension image data, this retrieval method obviously has been unable to meet efficiency. It is more important that the image database should be preprocessed and establish index to improve retrieval efficiency. As so, this paper introduces a non-parametric, efficient, subspace-based clustering algorithm for image database preprocessing. Meanwhile, the author improves the algorithm, and uses the multi-scale wavelet analysis technology to process the distribution curve of samples, and creates a multi-layer cluster. Based on the multi-layer cluster, it can establish hierarchical indexing structure. Moreover, the author proposes to use the database system to preserve this index structure.The theoretical basis of the above, the author designs a CBIR experimental platform. Compared by retrieval efficiency and retrieval result, it is convincingly proved that hierarchical index structure based on clustering is efficient and applicable in CBIR.