A fast indexing and retrieval method for image databases

Abstract

Image indexing is an important research issue nowadays. There are some ways to represent images such as 2D string. 2D string is a tool which represents spatial relationships among objects in symbolic image databases. 2D strings with respect to 3D hash functions are used in this paper to present a fast method for partial or total image retrieval. In the proposed method, a new data structure based on multi level linked lists is proposed to reduce algorithm time complexity. The proposed method overcame the constraint of other methods in image rotation. Because this method does not need to reconstruct data structures at new insertion, it can be extended toward dynamic databases. Simulation results showed the performance of our method. We concluded based on experimental results that our method time complexity tends to O(1).