Research on semantic network image retrieval method

摘要

As a prominent form of multimedia, image retrieval has become an important project research presently. Nowadays, the development of image search engine mainly bases on two kinds of technique: (1) traditional Text-Based Image Retrieval (TBIR); (2) Content-Based Image Retrieval (CBIR). However, because of the limitation of the ¿semantic gap¿ bottleneck, they both have limitations. In the light of this, we present an image retrieval method based on semantic network. We create a mapping from low-level image visual features to high-level semantic, and attempt to identify the semantic concept of visual features. We also introduce user feedback, guide search results to the optimal direction, and make it to fit the natural way for humans to understand image. The technology requires the use of the knowledge library for storing semantic networks and mapping. In this paper, the system model, retrieval method and experiments are given. Experimental results indicate that the method have better retrieval efficiency.