A Multimedia Database Management System Based On Face Recognition

For decades, face recognition has been an active topic in image processing and pattern recognition, which is theoretically important as well as urgently needed in many applications. So the application of face recognition to the field of the management of the personnel multimedia information and the requirement for video surveillance are aimed at, and a practical database management system is developed. For the system, the multimedia database is developed with SQL Server and the main body of the system with the programming language Visual C++.At first, technology of face recognition is summarized. And the common flows of face recognition and means used in every flow is presented emphatically.The second, some algorithms of digital image processing used in the system, i.e., the complexion model used for face detection and the n-tuple neural networks with sparse RAM (Sn-tuple) used for face recognition, is presented. Based on these, the address setting of sparse distributed memory (SDM) of Sn-tuple neural network model is researched, and two betterment means are brought forward, one is a random way that address setting range of SDM is reduced, and the other is a non-random way that SDM address matrix is made up of several pixel matrices of the trained face images. To prove this, a lot of experiments on this system are made and better performance of face recognition is gained.Lastly, the system function and ways to design every function models are presented emphatically. And the system performance in face recognition is tested by combining the practical face images database and the ORL face images database, and gained satisfied results.