In the content-based image retrieval technology, the performance of retrieval system using only a single image feature is generally unsatisfactory, and therefore the image retrieval system using two or more image features is more often used. When there is the target deformation or the size variation, the performance of image retrieval system using only shape features is not satisfactory, too. To solve these problems, in this paper, the extraction of image salient region and a shape representation methods of describing the image content are proposed, then they are used with image texture and color features to implement image retrieval. Experimental results show that the proposed image retrieval system can provide very good retrieval performance.