Study On The Small Human Facial Recognition Prototype System

The human facial recognition technology is the accepted kindness and in-constraint way than other biometric recognition method. From 1888, the first formal human facial recognition was put forward, to nowadays, many scholars and research departments have dedicated to the research of human facial recognition system. Some system was made for practicable purpose. Most of them, however, were run based on huge or parallel servers which cost much money and had big volume. It was so difficult to make it for commercial and ordinary purpose.A small human facial recognition prototype system is introduced. The system uses the digital signal processor to implement the mainly processing function and the micro control unit to achieve the whole control purpose. Moreover, a facial recognition algorithm is based on discrete cosine transform which was designed for the system. The hardware sub-system mainly include human face image capture unit which consist of general camera, digital video decoder and video port of the DSP, image display unit which be made up by digital video encoder and color LCD display, the store unit with SDRAM and FLASH memory, the serial communication link and the control unit which integrated with AT89C51, alphanumeric LCD and some buttons. All of the mentioned units are integrated into a whole system.A based on human face skin method for face detection approach is implement. For getting rid of the effect from the light, a self-adapt gray stretch procedure is applied to the face gray image, and then the face eigenvector are extracted from the data after DCT. The similitude of two face eigenvectors are calculated out by means of distance formulate. Finally, the system and algorithm are verified and the result are analyzed, some advice for promote purpose are presented.The prototype system of human facial recognition cost little money and takes a few of room and has the characteristic of high performance and flexible control ability. The algorithm of facial recognition is a quick and simple method and the face eigenvectors are occupied small memories. It is a feasible reference framework.