Face and speech recognition on field programmable gate array

The aim of this study is to develop a system by using Field Programmable Gate Array (FPGA) to recognize the people based on face and speech data.In the research stage of the study, feature extraction techniques and development strategies of FPGAs in the market were searched. Based on the selected recognition methods, recognition software was developed which intends to include the advantages of FPGAs in the area of biometric recognition. In this study, general procedure can be divided into training phase and recognition phase. In the training phase, face images and speech recordings were processed separately by using feature extraction techniques. Principal Component Analysis and Fourier Transform Analysis are used as feature extraction methods for face and speech recognition, respectively. The matrix which included all face images and speech recordings in database was processed in FPGA. Hence, a feature matrix was created and stored in the memory of FPGA. In the recognition phase, a face image and a speech recording was processed in FPGA to create a significant feature vector. The resultant data was compared to the related database to find the owner of the incoming data. This thesis also includes the development steps of a multibiometric recognition system which basically combines the face and speech recognition systems. It makes a fusion process at the decision levels of face and speech recognition systems.This thesis presents the results of systems in many cases and provides the possible advantages and disadvantages of multimodal recognition systems against unimodal recognition system.