Dynamic hierarchical semantic network based image retrieval using relevance feedback

In order to improve the retrieval accuracy of image retrieval, semantic-based image retrieval has become popular in the recent years. It is because this kind of retrieval method could narrow “semantic gap” between the visual features and the high-level semantic features. However, most of the existing methods of the semantic-based image retrieval are limited to fixed number of semantic features. A dynamic hierarchical semantic network method is proposed in this paper to overcome this limitation. The proposed dynamic hierarchical semantic network is constructed by relevance feedback. The number of semantic features can be dynamically changed according to user feedbacks. Moreover, the semantic features are allowed to have different levels of abstraction. In addition, the proposed method also integrates low-level visual feature-based image retrieval style, which could full use of the advantages of visual feature-based image retrieval and semantic-based image retrieval. Experimental results show that the proposed method achieves higher retrieval accuracy than fixed number of semantic feature method and only one level semantic feature method.