Face recognition implementation using blackfin microprocessor

• Main Authors: Comia, Andres P., Sia, Kenneth Paulo G., Tumambing, Robin Roland A., Villarante, Clarice Anne A., Ambata, Leonard U.
• Format: text
• Published: Animo Repository 2024
• Subjects: Human face recognition (Computer science); Embedded computer systems; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/12149
• Institution: De La Salle University

This study is all about the development of a standalone embedded face recognition system to be applied for the security and safety purposes of a small office. The group will work with the Principal Component Analysis algorithm, which will be implemented in a Blackfin ADSP-BF537 Development Board. MATLAB and Visual DSP++ were used as the coding environment. The face recognition process begins by capturing the image using the OV07725 CMOS sensor. Raw images will now pass through the Blackfin AV EZ-Extender to get to the development board where the images will be detected and compared to other existing data stored in the board’s Flash memory. When an output is reached, data will now transfer to the LCD for it to be printed. With the system accuracy set to 80%, the boot-up period takes approximately 19 secs while the face recognition process itself takes only up to 3 secs.