Automatic Image Annotation and Retrieval Using the Latent Dirichlet Allocation Model

Content-based image retrieval faces a vital problem, namely “semantic gap” that exists between low level features and semantic concept. In order to solve this problem, image automatic annotations that allow users to access a large image database with textual queries are put forward. In this paper, the main study concentrates on an automatic image annotation method based on vector quantization (VQ) algorithms and Latent Dirichlet Allocation (LDA) model. VQ is used as a clustering and condensing technique. LDA model is introduced from text based information retrieval, which is widely studied nowadays and proved to be effective on discrete data processing and demission reduction. The experiment is carried out on the platform of MATLAB and applied to the Corel database of 400 images. The presented experiment results demonstrate the effectiveness of the proposed method in its application of automatic image annotation and keyword based image retrieval.