During the past decades, face recognition has received significant attention and obained great achievement, which make it potentially one of the most successful applications in biometrics. in the field of public security, face recognition technology is becoming an emerging alternative. althrough face recognition is easily affected by environmental factors, it begins to protect our lives and relieve human labors, especially under some constrained conditions. under the circumstances, toward practical applications, this paper proposes a novel embedded near infraread (NIR) face recognition system based on multi-core computer techologies to bridge the gap between research and real scenarios. for real-time and high performance requirements, we proposed an algorithm based on local binary pattern (LBP) and local phase quantization (LPQ). experiments are conducted on a large database, which demonstrate the advantages of our system.