This paper presents an intelligent method of retrieving images with Chinese captions from an image database. We combine color, shape and spatial features of the image to index and measure the similarity of images. As a technical contribution, a Seed-Filling like algorithm that could extract the shape and spatial relationship feature of images is proposed. Due to the difficulty of determining how far objects separate, we use qualitative spatial relations to analyze object similarities. Also, the system is incorporated with a visual interface and a set of tools, which allows the users to express the query by specifying or sketching the images conveniently. Besides, our feedback learning mechanism enhances the precision of retrieval. Our experience shows that the system is able to retrieve image information efficiently by the proposed approaches.
International journal of computer processing of oriental languages 13(2), pp.131-159