Symbolic image indexing and retrieval by spatial similarity: A new approach based on multi-dimensional B+ tree

In recent years, researchers have paid attention to image retrieval increasingly. By increasing the image volumes in databases, symbolic image that were used as summaries for physical images, seemed a good approach. This images are retrieved with shorter time than physical images and are cost-effective. In this research we introduce a new approach for symbolic image indexing and retrieving by using multi-dimensional B⁺-tree. Spiral relations multi-dimensional indexing mechanism is used for image retrieval. Different methods were proposed for retrieving this kind of images. Among those, the best method from speed and inflexibility point of view was making one dimensional B⁺-tree by using TSR algorithm that understands three-dimensional relations between image components. This would help database systems to be intelligent, flexible and speedy. We draw a B⁺ - tree for all images in database by using the low level colour feature and create multi-dimonsional B⁺-tree by combining it with the B⁺-tree that we made it for symbolic images by using triangle spatial relations(TSR). Empirical results showed that the presented method would have many advantages in comparison to one-dimensional B⁺-tree method.