Integrated system of face recognition and sound localization for a smart door phone

This paper proposes a smart system using both face recognition and sound localization techniques to identify foreign faces through a door phone in a more efficient and accurate way. In particular, a visitor´s voice is used to localize the proper location of the speaker. The location information is then used to adjust the door phone camera position to frame the face in its field of view. In order to accurately locate the facial position of the visitor, 4 microphones are positioned in a cross configuration. Furthermore, face detection and face recognition algorithms as well as a wireless interface are used to recognize the visitor and send the information to the home owner´s phone. The entire control system is built using an FPGA chip and tested for actual use in door phone environments¹.