In order to introduce effectively text-retrieval technologies into content-based image retrieval,a novel method is presented. Firstly,according to the human visual feature and principle of the block truncation code (BTC),the image is divided into non-overlapped and equally-sized sub-blocks which can embody the texture and edge information of the original image. These blocks are defined and the sub-features are extracted,based on which the image is mapped into the text. Then,the text is chosen as key words and the technology of text-based retrieval is adopted for image retrieval. The content of the image and the text-based retrieval technology are considered,and experimental results have shown that the proposed method has sound and robust retrieval performance.