Abstract: This paper presents a context-sensitive media similarity search algorithm. One of the central problems regarding media search is the semantic gap between the low-level features computed automatically from media data and the human interpretation of them. This is because the notion of similarity is usually based on high-level abstraction but the low-level features do not sometimes reflect the human perception. Many media search algorithms have used the Minkowski metric to measure similarity between image pairs. However those functions cannot adequately capture the aspects of the characteristics of the human visual system as well as the nonlinear relationships in contextual information given by images in a collection. Our search algorithm tackles this problem by employing a similarity measure and a ranking strategy that reflect the nonlinearity of human perception and contextual information in a dataset. Similarity search in an image database based on this contextual information shows encouraging experimental results.
Abstract: In order to accelerate the similarity search in highdimensional database, we propose a new hierarchical indexing method. It is composed of offline and online phases. Our contribution concerns both phases. In the offline phase, after gathering the whole of the data in clusters and constructing a hierarchical index, the main originality of our contribution consists to develop a method to construct bounding forms of clusters to avoid overlapping. For the online phase, our idea improves considerably performances of similarity search. However, for this second phase, we have also developed an adapted search algorithm. Our method baptized NOHIS (Non-Overlapping Hierarchical Index Structure) use the Principal Direction Divisive Partitioning (PDDP) as algorithm of clustering. The principle of the PDDP is to divide data recursively into two sub-clusters; division is done by using the hyper-plane orthogonal to the principal direction derived from the covariance matrix and passing through the centroid of the cluster to divide. Data of each two sub-clusters obtained are including by a minimum bounding rectangle (MBR). The two MBRs are directed according to the principal direction. Consequently, the nonoverlapping between the two forms is assured. Experiments use databases containing image descriptors. Results show that the proposed method outperforms sequential scan and SRtree in processing k-nearest neighbors.