Retrieving Content Directly from Antique Book Images

The digitization and utilization of antique books are of significance. WWW lays a foundation for the dissemination of images, but more comprehensive mechanisms, such as content-based retrieval, are necessary to reduce information overload. It is well known that the content retrieval of a Chinese antique book directly from its bitmap is very hard due to intrinsic handwritten nature. In this paper, an original method for content retrieval based on visual similarity is proposed. By extracting features from book images, the method makes up a feature space and applies spatial indexing on it. A range searching strategy is then applied to retrieve all analogs to the query example. The prototypical implementation demonstrates the feasibility and supports both retrieving and browsing via popular Web browsers.