An album management system which can intelligently classify a large number of digital photos is present and designed to meet the need of the effective management of the growing number of digital photos. The key technologies include extracting the extra information of digital photos, face detection, face annotation, face recognition and feature matching, which are mainly implemented by OpenCV vision library. Experiments show that the system can effectively automatically implement the classification of time and people number, and semi-automatically implement the classification of people and scene by interactive method, which has a high application value.