Real time face matching with FPGA on video images

With the development of security systems, the use of biometric systems has become widespread today. Biometric systems are generally preferred for recognizing people's identities. Among the biometric systems, face recognition systems have become more widespread due to the simplicity and ease of use. In this research, local binary patterns (LBP), Eigenfaces, Fisherfaces algorithms used in traditional face recognition systems were examined. It is understood that the time efficiency decreases due to the high number of operations to be carried out at the same time in the analyzed algorithms.In this study, a new hardware-based accelerated face matching system was developed to calculate multiple processes at the same time, where the performance of commonly used computer processors remains low. For the proposed hardware-based algorithm, analysis was performed using the Nexys 4 DDR card containing FPGA from the Xilinx Artix-7 series.It has been shown that the proposed method has 5.7 times faster time saving.Since the improved Local Dual Pattern method was designed flexibly in a modular structure, it was found that it can also be applied to more advanced FPGA cards.The system has been tested using the ORL dataset. Two examples regarding the use of the developed method in daily life were applied and valid results were obtained