Fast Similarity Search in High Dimensional Image Data Sets

Abstract

Due to the development of images data production and use, the size of images databases is exponentially increased recently. Consequently, the similarity search time in images databases becoming a severe problem which affects the exploitation of image data. In this paper, we address the problem of reducing the similarity search cost in large, high-dimensional and dynamic image datasets; we propose a coherent improvement of the D-index method, the amelioration aims to reduce the number of distances computed during a query search. The method is tested over a 112 dimensional color images databases, the experiences show that the proposed D-index version has proved a good performance in comparison with the classical D-index.