Abstract：A digital library is a kind of information resource using the novel technology. An image retrieval based on texture features is proposed to solve the low efficiency in document retrieval for a digital library. Firstly, according to the characters of document images and by combining with the human visual, the edge information is extracted. Based on thusly obtained information, the significant texture feature is introduced. Then, the description is achieved and used in document image retrieval. The experimental results demonstrate that the proposed method has low complex and high precision in image retrieval for a digital library.

Keyword：digital library;content-based image retrieval (CBIR) ;document image;human visual;