Abstract: Semantic image retrieval has been a crux to bridge "semantic gap" between the simple visual features and the abundant semantics delivered by an image. The state of the art of semantic image retrieval is given from three aspects: schemes of image semantic description, approaches to image semantic processing, and design for image retrieval system. Firstly, a few of object oriented image content models and image semantic content representations are introduced. Secondly, some typical methods for semantics extraction are analyzed in detail by classifying them into three classes: extraction by knowledge, extraction by interaction, and extraction from exterior information. Thirdly, the user interface and different semantic processing methods in image retrieval systems are discussed. Finally, four critical problems faced in semantic image retrieval, including object recognition, semantic knowledge base, user retrieval model, and performance evaluation, are explained. Towards them, some resolving strategies are presented elementarily.